Sybase数据库技术，数据库恢复专家

2. Go to the Rights tab of the User prop sheet. Note that the Connect right has not been added to the user.

We should have a mechanism a checked-grayed box for inherited rights.

It could also be a "show inherited rights" icon like in member tab for a group.

BUG when creating repository tables. No confirmation message to say "Finished creating tables."

I know that this is not a major issue, but from the customer's perspective, they need to know when a PowerDesigner process has completed.

During consolidation of large models, PD shell stops responding to Windows (doesn't update display, etc). Need more than just "Saving the local document in the repository..."

1) Import model into PD7 and save it

2) Then consolidate that model into your favorite repository.

3) When it gives you the "Saving the local document..." message, launch the TaskManager and check out the PowerDesigner program.

NOTE: It will be listed as "Not responding".

4) If you have another application open, move its window around in front of the PD GUI.

Notice that it doesn't refresh the display. This is not good.

It takes a long time (anything over a minute or so is too long, but I'm talking 30 minutes and longer here) before it will update the GUI.

What I want to see is more GUI refreshes and some kind of status update in the message window, like what happens when I Reverse Engineer a database. The old MetaWorks gave me more information than this when it was working. Just giving me an hourglass is not acceptable

When creating the repository dictionary on DB2, there is no way to specify Create Table options that are allowed in DB2. Also, there is no way to obtain a copy of the DDL script for modification before the repository is created.
In PD6, there was a SHOW option in MetaWorks too provide access to the Dictionary creation script.

This functionality needs to added back into PowerDesigner 7.5.x.

In the PowerDesigner Repository Users Guide Part 1 "Using the Repository" chapter 7 topic "Understanding consolidation conflicts," there is a paragraph that reads as follows:

Updating PowerDesigner models

When you update PowerDesigner models, the merge dialog box
displays differences between models and lets you define merge actions.
In the merge dialog box, the local model is the source model displayed
in the right pane, and the repository model is the target model that will
be modified in the left pane of the merge dialog box.

Where it says that "the local model is the source model displayed in the right pane, and the repository model is the target model that will be modified in the left pane of the merge dialog box" THE OPPOSITE IS TRUE, THE TARGET MODEL WILL ALWAYS BE ON THE RIGHT SIDE AND THE SOURCE ON THE LEFT. [Passing along customer's upper-case emphasis there, that comes from confusion we had working with the Merge dialog in the Betas.]

The doc is wrong and needs to be updated because Source-Target is potentially confusing.

BUG in Repository Extract, Select Packages dialog box. The list of packages is NOT SORTED. With over 125 in a model, this doesn't work well. Also EXTREMELY SLOW getting just one package.

Allow the user to build folders by reading a directory and then to consolidate that into the repository, or go the other way.

We have all of our models organized into a fairly complex directory structure on a network server. I want to duplicate the hierarchy of the directories directly in PD. Then I want to duplicate that into the repository. Now, when I extract something from the repository, it should default to the correct folder in the workspace, and from the workspace it should automatically save to the proper directory on the server. This should be a General Option like "Keep all folders in sync".

It would be nice to send that SQL stream to a text LOG file, maybe put messages in the output window such as "Updating PMREFR table..." (one per table) or something to that effect. At the end, put a note that says "The entire SQL stream can be found in the file C:\Program Files\Sybase\PowerDesigner 8\RepositoryUpgrade.LOG" or something like that.

NOTE: The displaying of all of those statements (and I had thousands upon thousands) actually slows down the process significantly. Logging them to a text file could actually make it run faster.

Please provide the user with some guidelines as to how large to make the repository database (at least for starters). There could be recommendations for a small, medium, and large model. And all the disclaimers can still be documented.

Consolidation of a model into the repsoitory takes a significant time if number of entities is large (> 100). It would be very helpful to see a progress bar showing % of consolidation completed.

When you look at attributes of an entity in the Repository tree view, they all appear as "Entity Attribute" and their property sheets have Name and Code blank. The tree view should show the Name of the attribute, and the property sheet should have the Name and Code value for the attribute.

To Reporoduce (I used Sybase ASE 12.0 on NT for the Repository and build 203 on NT for PD8):

[ 1] Open PD8 Examples\project.cdm and consolidate into Repository.
[ 2] Tools -> General Options -> Repository check on "Show objects" in Browser group if not on already.
[ 3] In Repository Object Browser, expand model node and Entiities node under that. Node entries are shown for the entities in the model.
[ 4] Pick one -- for example, CUSTOMER and open its node. Open the attributes node under that. For CUSTOMER, there are six items in the tree list that say "Entity Attribute" -- these should show the attribute name.
[ 5] Open the property sheet for these items. They all have the Name and Code fields blank. They should contain the correct values for the Name and Code of the attribute.

There is no icon to open the property sheet of the parent object as you can do when you are in a CDM, PDM or OOM. The Property sheets for shortcuts in the Repository are not connected to the parent objects. For example on a shortcut entity, you cannot get the property sheet of the parent object. You cannot see its data item attributes, and the identification of the object's parent model and package is not clear. This functionality should be added.

Also, the General tab has a field called Location that has a value "Project management" -- the model name, telling you that this shortcut is in the model level of information. On the Attributes tab, there is a line entry also called Location that says "\CUSTOMER MANAGEMENT" -- reflecting the package where the entity resides. This is a little confusing. In the property sheet for shortcuts in the workspace, these fields are called "Target model" and "Target package" which is clearer.

Repro:
[ 1] Open PD8 Examples\project.cdm and convert the two submodel diagrams to packages.
[ 2] Consolidate into Repository (I used Sybase ASE 12.0 on NT, with Build 203 of PD8).
[ 3] Make sure "Show objects" is turned on in Tools -> General Options -> Repository -> Browser groupbox. Look at the tree view of the Repository Object Browser and see the shortcuts for entities at the model (top) level. Open the property sheet for one, for example, CUSTOMER.
[ 4] See that there is no way to get at information about the parent object.

The problem appears to only occur when composite primary keys are being used and the first column of the primary key has a distinct value less than 100%.

To reproduce:
1) Create a new PDM
2) Create two tables
3) In table_1, add two columns and make them both part of the primary key.
4) Change the Distinct Values to 40% on the first column of primary key (column_1)
5) In table_2, add two columns and make them both part of the primary key.
6) Change the Distinct Values to 40% on the second column of primary key (column_2)
7) Consolidate the PDM (Note: Do not allow the extraction of the PDM)
Compare the PDM in the repository with the PDM in the workspace.
Note: In table_1, the distinct value percentage of column_1 has been changed to 100% in the repository.

Enhancement Request: In the Merge window pane (upon consolidation): you must uncheck each specific Sybmol's checkbox, this could take alot of effort... Somehow offer the user the ability to be able to "DESELECT ALL SYMBOLS" for each diagram. And then go to a specific diagram and select all the sybmols within that Branch. Or have a Select All/Unselect all symbol objects in the Symbols branch. But, it would be nicer to be able to go to a diagram node under symbols and be able to select all/de-select all the symbol nodes in a specific diagram.

To reproduce the behavior:

a) User1 extracts the current version of the model to their workspace.
b) User2 extracts the current version of the model to their workspace.
c) User1 changes the fill color of entity1 symbol on diagram1 from white to blue.
d) User2 changes the fill color of entity2 symbol on diagram2 from white to red.
e) User1 consolidates their workspace to the repository. The repository recommends (by prefilling in the check box) that the local change for entity1 be applied. The extracted model reflects the new color of blue for entity1. Entity2 is (still) white.
f1) User2 consolidates their workspace to the repository. The repository recommends (by prefilling in the check box)that the local copy for entity2 be applied. The extracted model reflects the new color of blue for entity1.
f2) BUT the repository also recommends (by prefilling in the check box)that the local copy of entity1 be applied - even though user2 has never made any change to diagram1 or entity1. The extracted model reflects the old color of white for entity1.

Per Product Development:

In PD 8.x, currently, the changes to symbol's attributes (fill color, line style, background color, font style, font style etc...) are not updated by a modification date in the Version info tab on that object. The Modification Date only concerns the object itself (adding/deleting a column, change a data type, add in description, comments etc..). So merge during the consolidation does not know when the change has been done on the symbol and cannot propose a correct check box. So when you make a change on the symbol, merge proposes a default checked check box but it can be turned on to be the wrong result. The best way to solve this issue would be to manage a modification date on all symbols. Although, it will be memory consuming...

Repository Database : Informix 8 and 9

I reproduced on Informix 9 and ASA 7

======================================================

Steps to reproduce

- new pdm - 1 table

- consolidate and extract

- modify the PDM like that

Models --> Modeles properties --> Database --> New

add few properties : Database name, Comments, notes

- consolidate with update

- extract again, all Database properties still like before nothing has been updated.

Reproduced with PD 7.5, 8.01 build 289 and 9 build 437

Repository : ASA 7

Error message:

" Unexpected error while executing the following action:

Changing value of attribute Alias of object Query Column Object

The consolidation of the document has been canceled."

After modify a view in a pdm by adding a new alias or adding a column from a new table and then consolidating the model using update, results in an error:

Unexpected error while executing the following action:

Changing value of attribute Alias of object Query Column Object

There is no way to update the repository with a change made to the order of an object, send backwards or bring forwards. If this change is made, consolidating the model does not show any differences. If an object is modified so that the consolidation occurs, the change in object order is not updated to the repository.

Steps to repro;

1) create a new CDM
2) create 2 entities (entity_1 & entity_2)
3) position the symbols in such a way that entity_2 is overlapping entity_1 in a cascading fashion
4) consolidate the model (I used Oracle 8.1.7 rep using Merant 3.70 ODBC) using all the default settings
5) extract the model (do not update model in workspace, create a new model). You now have two models in your workspace with the same name
6) right-click on entity_1 >Order>Bring to front. Now it wil be overlapping entity_2. If you try to consolidate it at this point, no consolidation will occur becuase PD shows no differences between the local copy and the repository copy.
6a) You will need to make some other type of change to get the consolidation to complete. I changed entity_1's name & code to entity_10.
7) consolidate the model using all the default settings (update)
Extract the model and again do not update the workspace model. This will create a 3rd model in the workspace all having the same name.

NOTE:The name & code change remains intact however entity_2 is now in front of entity_10. The change made to the symbol was not recorded in the repository. PD is ignoring the object "order" information.

One of the bullets in this section states the following:

The Name of the folder. This name must be unique in a hierarchy of folders.

However, when actually naming the folders, the name must only be unique at each level of a hierarchy. For example, I can create a hierarchy of folder_1>folder_1>folder_1. (This makes the above bullet incorrect.) However, if I try to create a second folder_1 at the root level, I will receive a message stating that this name already exists.

A suggested description could be ....

The Name of the folder. This name must be unique at each level in a hierarchy of folders.
The title of the CR should read....
Please clarify the Using Folders section of the repository documentation.

background information:

User is trying to perform a visual comparison between a consolidated procedure and one in the local model. Based on this comparison, they would then like to copy sections of the consolidate procedure and paste them into the local procedure. (The way to accomplish the customer's comparison issue is to compare the two models, but the customer does not prefer the option.)

('rsinssys.sql' on Windows NT) contains the non-localized date value

'Jan 1, 4000' (and is used to represent a maximum date value). When the

ASE hosting the RSSD is using a non-English default language such as

Chinese, the date value is considered an illegal value and prevents the

proper initialization of the RSSD. To work around this issue, edit the

installation script and replace 'Jan 1, 4000' with '01/01/4000'.

greater than 127 unichars (or 254 bytes) in length, the Replication

Server terminates with the following messages in the error log followed

by a stack trace:

N. 2002/05/22 15:08:19. INTERNAL ERROR #20020 DSI EXEC(103(1)

PDS.pubs2_rep) - (7958) The packer failed to pack a command.

F. 2002/05/22 15:08:19. FATAL ERROR #1029 DSI EXEC(103(1)

PDS.pubs2_rep) - (7958) CS-Library error: Error: 33816868,

Severity 1 -- 'cs_convert: cslib user api layer: common library

error: The result is truncated because the conversion/operation

resulted in overflow.'.

To work around this issue, limit the length of a 'unichar' or

'univarchar' column to 127 unichars (or 254 bytes).

replicated databases

dbcc gettrunc can return two different column headings. One is ltm_trunc_state, as the doc mentions, and the other is secondary trunc state. If the value is 1 under either heading, then replication is enabled for that database.

Regarding subscriptions, the paragraph is confusing. Using the check subscription is a good idea for checking for "valid" subscriptions. Explaining about "create subscription" is confusing.

In the bold note in the first part of this section, we state that it is NOT sufficient to have data in the inbound queue of Replication Server. Yet, if the customer follows the instructions in this order, suspending the DSI connections before ensuring the transaction logs have been drained will put data in the inbound queue. The section for suspending the DSI connections should be after the section on draining the transaction logs.

Also, formatting wise, some of the 2>go commands appear on the same line as >1

Status: Reviewed Sep 2002. Fix targeted (but not commited) for the 12.6 release.

differing character sets, the Replication Server relies on the replicate

ASE to perform the necessary character set conversion on the replicate

data.

If the replicate ASE is configured to always use Unilib character set

conversion (the 'enable unicode conversion' configuration value is set

to '2') but configured with an insufficient 'size of unilib cache'

value, the ASE will issue the following warning and will not perform any

subsequent character set conversion.

Msg 2420, Level 17, State 1: Unable to load Unicode-based

character set conversion table 'sjis' - Unicode conversion cache

is full. Please notify your system administrator to add mode

unicode buffers.

The Replication Server ignores this ASE warning message, which results

in replicate data not undergoing the proper character set conversion.

To work around this issue, configure the ASE with a larger 'size of

unilib cache' configuration value. Validate that the new configuration

value is sufficient by confirming that the above-mentioned error message

does not appear when contacting the ASE using the 'isql'. Be sure to

invoke 'isql' with the appropriate '-J<charset>' parameters so that

'isql' uses the same character set as the Replication Server.

but primary table does

not exist in database.

Status: Reviewed Sep 2002. Fix targeted (but not commited) for a release beyond the 12.6 release.

"No to All" when the Installer asks about installing over existing

files.

If the Installer attempts to install a new file over an existing file,

the Installer will ask the user how to handle the conflict. If the user

choses "No to All", critical files will not get installed.

When the user attempts to start 'rs_init', 'rs_init' will terminate with

the a message similar to the following:

INTERNAL ERROR: Unable to load messages for product 'rs' using

language 'english' and character set 'cp850: could not locate

the localization file

'c:\sybase\locales\english\cp850\sybinit\rs.loc'.

To avoid this issue, install the Replication Server in a separate

directory.

directory as an older version of the Replication Server

If the Replication Server is installed in the same directory as an older

version of the Replication Server, critical files (e.g. 'libunic.dll')

will not be updated.

Without these updated critical files, the Replication Server will fail

to start and issue a message similar to the following:

Cannot find Procedure entry point unisort_DataCompare in Dynamic

link library libunic.dll.

To avoid this issue, install the Replication Server in a separate

directory from other existing Sybase products.

Additionally, it may be necessary to update the 'libunic.dll' file

manually.

To do this, copy the file ...

%SYBASE%\SYSDLL\WindowsNT\WinNT\system32\libunic.dll

... to the following location ...

%SYBASE%\REP-12_5\bin\libunic.dll

If this does not resolve the issue, copy the file ...

%SYBASE%\SYSDLL\WindowsNT\WinNT\system32\libunic.dll

... to the following location ...

%SystemRoot%\system32\libunic.dll

true if dump/load option is not selected.

Final screen of "add standby database wizard" should include following information:

If you choose to Initialize standby database using dump and load option, RSM will

not create the maint_user for you. See RS Admin Guide, "Adding the Standby Database

Maintenance User"

Status: Reviewed Sep 2002. Fix targeted (but not commited) for the 12.6 release.

negative. To fix this required changing the

schema of the XLog.

If you are creating an XLog, the new schema

will be used.

If you have an existing XLog with the new

schema (12.1.0.xxx where xxx >= 215 or

12.5.0.yyy where yyy >= 302), you need do

nothing special.

If you have an existing XLog without the new

schema (12.0 or 12.1.0.xxx where xxx < 215 or

12.5.0.yyy where yyy < 302), you must do the

following to migrate the schema:

1 - login to the RepAgent and verify that

scripts run automatically:

ra_config pdb_auto_run_scripts, true

3 - shut down the RepAgent and apply this EBF

3 - start the RepAgent in ADMIN mode

4 - lock the primary database to prevent the

marked primary tables and/or the marked primary

procedures from calling the ra_get_tx_info_

procedure

6 - at the RepAgent login, issue the pdb_xlog

command with no parameters

7 - unlock the primary database

8 - resume replication

The migration adds the ra_txid_seq_ table that

maps Informix transaction-id-related

information into a sequence number,

drops/re-creates the ra_get_tx_info_ procedure

to use the new table, and increments the

database generation id. Some additional

housekeeping entries are added to the

ra_xlog_system_ table, and, for Informix 7, the

tran-id rollover procedure (ra_txid_rollover_)

is created.

This new schema uses the following global

variables:

ra_begstamp_ (new)

ra_connected_ (new)

ra_opid_hi_

ra_opid_lo_

ra_pid_ (new)

ra_tstamp_

ra_txid_

ra_txid_hi_ (new)

ra_txid_lo_ (new)

Note that this fix does NOT migrate an XLog

created on Informix 7 to the appropriate schema

for Informix 9.

on start up.

causes a ClassCastException. This fix changes

the schema of the XLog and requires migrating

the XLog (see below). This fix also certifies

RepAgent 12.5 against Oracle 9.2, and supports

the Oracle TIMESTAMP and INTERVAL data types.

IMPORTANT: It is NOT recommended that you use

the Oracle 9.2 JDBC driver against Oracle 8i

because it causes the loss of time precision

from DATE columns in the XLog tables. Instead,

for Oracle 8i, use the Oracle 9.0 JDBC driver.

The following drivers have been certified:

for Oracle 8i, use

9.0.1.4 JDBC thin driver for

JDK 1.2 & 1.3 (classes12.zip)

for Oracle 9i (optionally for 8i), use

9.2.0.5 JDBC thin driver for

JDK 1.2 & 1.3 (classes12.zip)

MIGRATING the XLog:

If you are creating an XLog, the new schema

will be used.

If you have an existing XLog, you must do the

following to migrate the schema:

1 - login to the RepAgent and verify that

scripts run automatically:

ra_config pdb_auto_run_scripts, true

2 - shut down the RepAgent and apply this EBF

3 - if replicating out of Oracle 9i, upgrade

your JDBC driver to 9.2.0.5

4 - start the RepAgent in ADMIN mode

5 - login to the RepAgent and issue the

pdb_xlog command with no parameters

When the command returns successfully,

migration will have been completed and you may

resume replication.

Note that, regardless of how many fixes in this

release require migration, you need migrate a

given RepAgent instance only once.

replication envionment, 'rs_init' reports the following OCS errors ...

CONNECTIVITY ERROR: CT-Library error: 'ct_connect(): directory

service layer: internal directory control layer error:

Requested server name not found.'.

CONNECTIVITY ERROR: Login attempt failed: server = ''.

Despite this error, the database appears to be correctly added to the

replication environment.

operations from shadow tables and tranlog table

will fail where opid_lo > 2147483647.

(pdb_update_stats_type and pdb_update_stats_count)

are reset to default values at quiesce.

the ra_txid_seq_ table if it's busy.

Request to change topic "No RSM Client, no new features at source Replication Server"
to "No RSM Client, new features not yet implimented at source Replication Server"

Current wording could be misinterpreted that no new features were ever going to be used.

_dsi_qm_free_cmd -> exc_unlink_ele

Status: Reviewed Nov 2002. Fix to be targeted (but not committed) to the 12.5.1 release.

set identity_update <table name> on

... will not replicate properly when the table's identity column is

updated with the T-SQL UPDATE command ...

update <table name> set <identity column> = <identity value>

Under these circumstances, the Replication Server will ignore any change

to the identity column.

Version: PD 9.5 DBMS is Oracle 8i2, DB2 and ASA8 (This has been reproduce on each DBMS)

Steps to reproduce:

1. In PD 9.0 consolidate a PDM into the repository. (I used project.pdm)

2. Upgrade to PD 9.5 (using the web download)

3. Connect to the repository and run the update tables script. (connecting with ADMIN and DBA rights)

4. Extract the PDM model in the repository into PD 9.5

5. Make a change and consolidate it. (I changed the data type of a domain and added a table)

6. The following error message will occur:

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object Group 'PUBLIC'

The consolidation of the document has been

'writetext' command, and the TEXT/IMAGE column does not appear in any

replication definition at the Replication Server, the Replication Server

DSI will suspend with Error #32021 ("Function 'rs_datarow_for_writetext'

is not defined for table '<table>'").

'resume connection ... skip transaction' to skip a transaction that

contains modifications to TEXT/IMAGE columns. If a stack trace does

occur, the stack trace will contain the routine name

'_dsi_gen_textptr_init_output_cmds'.

For example, as an extension to autocorrection, per a replication definition, eliminate the command rather than delete/insert the update.

to an ASE is broken unexpectedly, the

Replication Server will attempt to

report the event. However, due to a

defect in the Replication Server, the

event is not properly processed and the

Replication Server will terminate

unexpectedly with internal error #12025.

Repserver can be configured to send encrypted passwords for out-going connections.

RA_PROCACTIVE_ is varchar(30) should be 61.

This fix changes the schema of the XLog and

requires migrating the XLog.

MIGRATING the XLog:

If you are creating an XLog, the new schema

will be used.

If you have an existing XLog, you must do the

following to migrate the schema:

1 - login to the RepAgent and verify that

scripts run automatically:

ra_config pdb_auto_run_scripts, true

2 - shut down the RepAgent and apply this EBF

3 - start the RepAgent in ADMIN mode

4 - login to the RepAgent and issue the

pdb_xlog command with no parameters

When the command returns successfully,

migration will have been completed and you may

resume replication.

Note that, regardless of how many fixes in this

release require migration, you need migrate a

given RepAgent instance only once.

set to NONE, SETUP00 must contain SYSADM parameter.

ability already exists in RepAgent to set the

various connections' charsets except for

RepServer's. So, rs_charset has been added as

a configurable property.

NOTE that, just because RepAgent provides a

pds_charset property, it does not mean that the

charset will actually be set at the vendor's

JDBC driver. Most vendor JDBC drivers provide

no way of setting the charset.

Feature request to return message if RSSD log full to RS.log to help diagnose RS hang problems.

ERROR #44048 dSERVICE( ) - /sysproc.c(189)
Error occurred while waiting for a child process.
Exiting due to a fatal error

Using the Data Direct 4.1 DB2 Wire Protocol driver in 9.5.1 fails with the following error:

[DataDirect][ODBC DB2 Wire Protocol driver][DB2]L964 SQLSTATE = S1000

command to a 12.5.0.3 ASE ...

set identity_insert ...

... the 12.5.0.3 ASE returns the following new informational (i.e.

severity 10) message to the Replication Server DSI ...

Message: 11065, State 1, Severity 10

''SET IDENTITY_INSERT' is ON for table '<table>'.

The Replication Server DSI will stop when encountering this

informational message because, by default, 12.5 Replication Server (as

well as earlier versions) does not define an error action for this new

informational message. If no error action is defined for any message,

informational or otherwise, the default action for the Replication

Server DSI is to stop.

Workaround (prior to EBF 10867) is to substitute unprintable chars for X'40' (space) or any other printable symbols.

This is also problem for any/all reserved words replicated by RS: insert, update, delete... Not just 'create table'

Workaround is to create repdef for standby table do not include timestamp as part of primary key set

replication with a syntax error.

Table PMLBIN and Table PMXFIL

Datatype Long Raw

So to be able to replicate all columns datatype, long raw has to be replace by : LOBs (CLOBs).

tries to break its connection while Repserver waits for the

completion of some tasks, an attention event can bring Repserver

back to handler the disconnection event. This may prevent Rep

Agent or RSI thread hang.

"Changing relative position of object table 'data_source'

linked to physical diagram"

Upon investigation in the merge screen, this table and a few others are being added to the 'symbols' section of the model. All the metadata properties, name, column,etc... is listed under symbols. This should not be possible.

Steps to reproduce the problem:

1. Make a connection to the repository I have created on conlabtt using the customers repository dump or restore dump from attached zip.

2. Repository login is xxxx and database info is Username: xxxx, password: xxxx

3. Extract the model PDM_MPVAR_test.pdm

4. Consolidate(update) it.

The following is the error message you will get:

Consolidating the document 'E:\program files\sybase\PowerDesigner 9\MPVaR_test.pdm'

- Extracting the repository document 'PhysicalDataModel_MPVaR'...

Updating model version from 9.5.0.660 to 9.5.1.759

- Comparing the local model with the repository model...

- Updating the repository document...

Create Physical Diagram "PhysicalDiagram_1" in the Package "testpackage"

Create Package "testpackage" in the Physical Data Model "PhysicalDataModel_MPVaR"

Create Table "Table_1" in the Package "testpackage"

Create Table "Table_2" in the Package "testpackage"

Create Table "Table_3" in the Package "testpackage"

Create Table "Table_4" in the Physical Data Model "PhysicalDataModel_MPVaR"

Create Package Symbol "testpackage" in the Physical Diagram "PhysicalModel_ALL"

Create Table Symbol "Table_1" in the Physical Diagram "PhysicalDiagram_1"

Create Table Symbol "Table_2" in the Physical Diagram "PhysicalDiagram_1"

Create Table Symbol "Table_3" in the Physical Diagram "PhysicalDiagram_1"

Unexpected error while executing the following action:

Changing relative position of object Table 'data_source' linked to Physical Diagram

The consolidation of the document has been canceled.

Note: The file does extract to the appropriate file type as specified in CR 216181.

To reproduce:

1) open project.pdm

2) select File\Save As and save as project2.pdm (binary file type)

3) select File\Save As a second time and note that binary is now the default type

4) consolidate project2.pdm

5) in the repository browser, right click on project2 and select extract

6) open the "save as" window

Note: The default file type is xml. It should be binary based upon the changes made in cr 216181.

repdef but not in the primary table AND

RepAgent column_compression is on. RepAgent

now adds the column (with a value of null) to

the LTL.

E. 2003/01/29 22:06:01. ERROR #60036 GLOBAL RS(GLOBAL RS) -

xec/exec.c(115)

Failed to acquire lock on sync element

'rsinfo_global_rshandle_mutex'. Info 8.

E. 2003/01/29 22:06:01. ERROR #60027 GLOBAL RS(GLOBAL RS) -

rshandle.c(390)

The Rep Server failed in an attempt to unlock the mutex for sync

element rsinfo_global_rshandle_mutex. (392)/servmain/rshandle.c.

E. 2003/01/29 22:06:01. ERROR #12020 GLOBAL RS(GLOBAL RS) -

/exchand.c(155)

Internal error. The free routine for resource type '2' did not

unlink a resource allocated at 'ic/useful/interapp.c(103)'

F. 2003/01/29 22:06:01. FATAL ERROR #18028 GLOBAL RS(GLOBAL RS) -

servmain.c(291)

Initialization failed. Server is shutting down.

From these messages it appears that a client appliction is attempting to connect to a RepServer that is in the middle of starting up. The attempt must have timed out, thus the return code of 8 from srv_lockmutex() - "attention received while waiting for lock". But as a result, RepServer's exception handler, assuming that the lock is held and needs to be released, tries to unlock it. This causes a chain of events that ultimately crash RepServer.

While notification of the failed login attempt needs to be retained, RepServer should not crash in these situations.

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object View Column

The consolidation of the document has been canceled.

The repository is installed on ASE 12.5

In PD 9.0.0, the repository users were added without passwords. When upgrading to PD 9.5.1, the repository upgrade fails.

If the user has a password in PD 9.0.0, then the repository upgrade will complete.

This appears to an issue with ASE only. When I followed the same steps with ASA8, the repository upgraded for each scenario.

To reproduce:

In PD 9.0

1) Create a repository with ASE as the DBMS

2) Add two users and assign administrator rights to both

user: sa password: <blank>

user:sapass password: sql

3) Close PD 9.0

In PD 9.5.1

4) Connect to the repository with user 'sa'

Note: you will receive the error messages previously mentioned

5) Connect to the repository with user 'sapass'

Note: the upgrade process proceeds as expected.

limits from 8KB to 16KB.

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object View Column

The consolidation of the document has been canceled.

To reproduce:

1) Consolidate model CID1.PDM. Have the extract after consolidation check box checked.

2) Open CID2.PDM and merge into CID1.PDM. Select all merge actions.

3) Consolidate(update) CID1. PDM again. Have the extract after consolidation check box checked.

4) Close and detach CID2 from the workspace.

5) Open CID3.PDM and merge into CID1.PDM. Select all merge actions and manually synchronize the diagrams, CORPORATE_DATABASE with M_CID_DB. The process succeeds but has three errors with incomplete joins (reported on CR 309417)

6) Consolidate CID1 in update mode.

7) Close and detach CID3 from the workspace.

Open CID31 and merge into CID1. Select all merge actions.

Note: This merge corrects the join criteria for the references in step 5

9) Consolidate CID1.

10)In the merge dialog box, Select all merge actions

Note: That the view object may be included eventhough it has not been modified. Also, if it is present, check the query information. I have been receiving two entries where one contains GUID information.

11) The consolidation fails.

The error text is:

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object View Column

The consolidation of the document has been canceled.

if the ASA connection went down

Use my change and keep repositoy change like that in merge window, object will be check or not in the repository part.

correct when column name has spaces.

left and right brackets instead of double

quotes when delimiting identifiers in create

and mark scripts. This allows identifiers to be

delimited, even when the SQL Server

QUOTED_IDENTIFIERS option is not set. Note that

you must still use double quotes when

delimiting an identifier in a RepAgent command.

automatically.

The function string is:

rs_dsi_check_thread_lock

the query is:

select count(*) 'seq' from master..syslocks where id in (select id from master..syslocks where spid = @@spid) and spid != @@spid

When there are conflicts, this query should return the number of threads that have outstanding (INTENT) requests for locks held by the thread.

However, when isolationlevel_3 is used, it appears the (INTENT) requirements are not displayed in the master..syslocks table.

Normal isolation:

1> sp_lock

2> go

The class column will display the cursor name for locks associated with a cursor

for the current user and the cursor id for other users.

fid spid loid locktype table_id page

row dbname class

context

------ ------ ----------- ---------------------------- ----------- -----------

------ --------------- ------------------------------

----------------------------

0 1 2 Sh_intent 448004627 0

0 master Non Cursor Lock

0 15 30 Ex_intent 2032010270 0

0 pdb1 Non Cursor Lock

0 15 30 Ex_page-blk 2032010270 770

0 pdb1 Non Cursor Lock

0 26 52 Ex_intent 2032010270 0

0 pdb1 Non Cursor Lock

(4 rows affected)

With isolationlevel_3

1> sp_lock

2> go

The class column will display the cursor name for locks associated with a cursor

for the current user and the cursor id for other users.

fid spid loid locktype table_id page

row dbname class

context

------ ------ ----------- ---------------------------- ----------- -----------

------ --------------- ------------------------------

----------------------------

0 1 2 Sh_intent 448004627 0

0 master Non Cursor Lock

0 15 30 Ex_intent 2032010270 0

0 pdb1 Non Cursor Lock

0 15 30 Ex_table-blk 2032010270 0

0 pdb1 Non Cursor Lock

(4 rows affected)

adding the standby database for a warm standby system.

T. 2003/03/14 17:17:07. (230): Thread DSI(193 DS_STANDBY.ops) infected

with signal 10

...

T. 2003/03/14 17:17:08. (230): ***************STACK TRACE***************

T. 2003/03/14 17:17:08. (230): *****thread DSI(193 DS_STANDBY.ops)*******

T. 2003/03/14 17:17:08. (230): pc: 0xfa228, dbg_dump_stack ()

T. 2003/03/14 17:17:08. (230): pc: 0x748d8, exc__appDumpStack ()

T. 2003/03/14 17:17:08. (230): pc: 0x75650, exc_terminate ()

T. 2003/03/14 17:17:08. (230): pc: 0xba260, bsd_catch_signal ()

T. 2003/03/14 17:17:08. (230): pc: 0xb9b54, sys_init_signals ()

T. 2003/03/14 17:17:09. (230): pc: 0x48040, ll_ins_after ()

T. 2003/03/14 17:17:09. (230): pc: 0x3c8fc, exc_unlink_ele ()

T. 2003/03/14 17:17:09. (230): pc: 0x4fef0, _sync_cancel_sleep ()

T. 2003/03/14 17:17:09. (230): pc: 0xdce54, sqt_LoadCache ()

T. 2003/03/14 17:17:09. (230): pc: 0x29b00, dsi__SchedulerBody ()

T. 2003/03/14 17:17:09. (230): pc: 0x296e8, dsi__SchedulerWrapper ()

And another:

T. 2003/03/15 11:15:24. (27): Thread DSI(193 DS_STANDBY.ops) infected with

signal 10

...

T. 2003/03/15 11:15:25. (27): ***************STACK TRACE***************

T. 2003/03/15 11:15:25. (27): *****thread DSI(193 DS_STANDBY.ops)*******

T. 2003/03/15 11:15:25. (27): pc: 0xfa228, dbg_dump_stack ()

T. 2003/03/15 11:15:26. (27): pc: 0x748d8, exc__appDumpStack ()

T. 2003/03/15 11:15:26. (27): pc: 0x75650, exc_terminate ()

T. 2003/03/15 11:15:26. (27): pc: 0xba260, bsd_catch_signal ()

T. 2003/03/15 11:15:26. (27): pc: 0xb9b54, sys_init_signals ()

T. 2003/03/15 11:15:26. (27): pc: 0x4804c, ll_del ()

T. 2003/03/15 11:15:26. (27): pc: 0x74cb4, _exc_alloc_element ()

T. 2003/03/15 11:15:26. (27): pc: 0x3c7f8, exc_link_ele ()

T. 2003/03/15 11:15:26. (27): pc: 0x4fe94, _sync_will_sleep ()

T. 2003/03/15 11:15:26. (27): pc: 0xdcf0c, sqt_LoadCache ()

T. 2003/03/15 11:15:26. (27): pc: 0x29b00, dsi__SchedulerBody ()

T. 2003/03/15 11:15:26. (27): pc: 0x296e8, dsi__SchedulerWrapper ()

T. 2003/03/15 11:15:26. (27): pc: 0x291a8, _dsi_qm_cmd_retry ()

T. 2003/03/15 11:15:26. (27): pc: 0xc3195838, signal ()

T. 2003/03/15 11:15:26. (27): pc: 0xc0044afc, signal ()

T. 2003/03/15 11:15:26. (27): *****End of stack trace.****

T. 2003/03/15 11:15:26. (27): Internal error. Attempting to produce a core

file

PD 9.5.2 build 856 : fixed

Consolidation with update does not preserve an operation associated to a message inside a sequence diagram when Actor and object are shortcuts.

This is reproducable only when repository tables reside on a Ms Sql Server 2000 database.

NOTE: I cannot reproduce on Sybase ASE 12.5, ASA 7.0.x, ASA8.0.x or Oracle 8.1.7 dbms...

Tested using either of the following ODBC drivers:

Microsoft Sql Server v. 2000.81.9001.00 SQLSRV32.dll 2/19/2002

PB DataDirect OEM 4.10 32-BIT Sql Server Wire Protocol v. 4.10.00.00 PBMSSS18.dll 2/15/2002

PB DataDirect OEM 4.10 32-BIT Sql Server v. 4.10.00.00 PBSS618.dll 2/15/2002

To reproduce:

1. Consolidate (no auto extract) the sample "project.cdm" file to a Ms Sql Server 2000 database.

2. In Repository browser, select the Entity folder > select a specified entity and expand it's tree. You will see entity attributes listed as "Entity Attribute" and not it's given name.

3. Double click a entity attribute that is displayed as "Entity Attribute" to get it's properties window. You will see it's Name and Code entries are empty.

NOTE: it appears to only be a display issue in the Repository. Because, when you extract the model and view it in the workspace window, the list of entity attributes are displayed and listed correctly.

Option to only replicate DDL and not DML for WarmStandby.

something like...

sp_reptostandby <PDB>, 'DDL only'

or have sp_setreptable 'false' prevent tables from replicating

Document in the "Starting RS" Chapter of Admin Guide... Guidelines for Adding RS Domains.

Specifically add ASA custom function string rs_triggers_reset to allow set triggers off option. Default rs_function_string_class rs_triggers_reset does not work for ASA.

When this option is set, the PC using 100% of the CPU and performance is affected on the server where other application cannot access other dbms on same server.

The customer has installed the repository on ASA 9 and ASE 12.5 with the same results.

PSE has tested on ASA 8 with the same results.

To reproduce:

1) Consolidate the customer's models into a repository.

The models inlcude a LDM which was then used to generate three PDMs with different target DBMS. The files are 2.8, 4.3, 4.8, and 5.3 MB in size.

2) Extract the LDM, 2.8 MB, it takes 47 seconds.

3) Extract the same model, LDM, and select the 'extract dependencies' option.

Note: It takes about one hour to extract the model

as much as JRE 1.2.2 supports.

The where clause colname is not listed by rs_helpsub

"WARNING: unexpected result (2) from comparing two previously inconsistent rows.

[c/subcmp.c(1792)]"

when comparing univarbinary/univarchar datatypes.

longer logged in the system.log because they

are (trace) informational only and fill up the

log unnecessarily:

E. 2003/04/15 14:55:12.64.

com.sybase.ra.conn.query.ColumnMetaData

Attempt to retrieve column precision for

column of SQLType <2004> failed.

<NumberFormatException> thrown by JDBC

Driver. Precision being set to 0.

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object Shortcut of Column 'ENTY_PRFL_ID'

does not work.

empty_clob() does not replicate.

listening client activated. The parameter setup for message persistency is needed before

sending out the JMS messages to the assigned JMS server that doesn't implement the

default persistency of J2EE standard.

Full, Write, Read and List.

Customers would like to have a new option to Visualize or Not the model in the future, without dropping the User Group for that model.

Today, it's possible to disallow some user to visualize a model by dropping his user group (PUBLIC (All Users), for example) from that model, only.

definitions per replicated stored procedure.

FOR EXAMPLE:

create function replication definition <name>

with primary at <dsname>.<dbname>

with all procedures named <sproc name>

"with all procedures named" would allow function definition name

to be different than the procedure name. Therefor, you would then

be able to create more than one function replication definition

for that one replicated stored procedure.

Currently, the function definition name must be the same as the

pdb.procedure name... and customers must create multiple procedures,

each with their own function replication definition.

includes the RSSD version as well as more

complete jConnect version info.

"resume connection skip transaction" command.

replicating a complex stored procedure.

Usage section: Use sysadmin dump_queue to drop the contents of a Replication Server stable queue.

This should be: dump the contents of the queue!

which is requesting the additional memory which request can't be satisfied

To reproduce:

1) Create a repository definition to an ASA 8 repository

2) Select the General Options > Repository > Auto-connect option

3) Close PD

4) In Windows explorer, either rename or move the database file listed in the ODBC DSN used for the repository connection

5) Open PowerDesigner

NOTE: No warning messages and no connection to the repository

6) Select Repository > Connect

Note: An error message is generated

with the persistency of messaging for multiple subscribers.

[var]char/[var]binary data that is longer than

255 for RepServer 12.1 and earlier.

If you want the repdefs generated by RepAgent to use the heterogeneous datatype definitions defined in RepServer, set the lti_use_hdds parameter to true before either issuing the lti_gen_repdefs command or before resuming replication (this last assumes that you have already set lti_auto_gen_repdefs to true).

that is from a Repserver 'system' tables point of view

check references/inconsitencies amongst them.

as 0x1f3f when RepServer is cp850. For more

details, see description of fix for CR #349444.

and immediately terminates Replication Server.

Right now the upgrade is too cumbersome to apply to a production system. See page 52 of the Installation and Configuration Guide for 4.1 EBF 2

calls with a traceflag or 'configure connection' option could increase RS performance...

charset setting and conversion. For more

details, see description of fix to CR #349444.

A route version should reflect the lower value between the site version of your primary Replication Server and the site version of your replicated Replication Server. For example, if the site version of your primary Replication Server is 12.1 and the site version of your replicated Replication Server is 12.0, the route version should be 12.0.

However, the version of rs_fast_route_upgrade provided in Replication Server version 12.1 and earlier erroneously sets the route version to the higher value. For this reason, the earlier version of rs_fast_route_upgrade automatically upgrades the route version to 1260 even if your primary Replication Server has not yet been upgraded to version 12.6.

If this applies to your replication system and you are upgrading to Replication Server version 12.6, the upgrade script stops and tells you to fix the route version before upgrading to version 12.6.

The Sybase Log Extract (called only from 12.6 if LogExtractor=LTMEXT) does not return the proper return code when it detects the load libraries are not authorized (Authorization Program Facility or APF), it returns a pointer to the return code instead of the value 8.

"AS" keyword in a stored procedure, only the

"IS" keyword.

qualified identifier in DDL. If MSA is used to replicate this DDL and

the replicate DB does not have the same name as the primary DB, MSA

will not be able to apply the DDL and DSI will be suspended.

create table datetime

(row_num int,

datecol date,

timecol time,

ndatecol date null,

ntimecol time null,

comment varchar(50) null)

When repserver tries to query from the table at de-materialization, repserver gets the following error message:

Message from server: Message: 247, State 2, Severity 16 -- 'Arithmetic overflow during implicit conversion of DATE value 'Jul 7 *''*' to a DATETIME field .

Workaround solution is to drop subscription without purge then manutally purge the replicate table.

When using the traditional ASE RSSD, these conversions are processed and displayed correctly.

property pdb_convert_datetime is true, then a

DATETIME field may not get converted to Sybase

format correctly.

You'll get the error message:

There is already a model that has the name ... in the folder ...

PD uses the old foldername and it is not possible the foldername for Integrate.

This may be related to Cr 295333

To reproduce:

1) Consolidate customer's models, cdm_shrtct1.cdm, ProMan_target1.cdm, and pdm_shrtct1.pdm, without extraction

2) Extract customer files to a different folder, repos, or as a different file name

3) For the repos\cdm_shrtct1.cdm, select Target Model and change the files to those extracted to the repos folder, repos\ProMan_target1.cdm, and repos\pdm_shrtct1.pdm, and save the model

4) For repos\pdm_shrtct1.pdm, select Target Model and change the file to the one extracted to the repos folder, repos\cdm_shrtct1.cdm, and save the model

5) Consolidate repos\cdm_shrtct1.cdm and repos\pdm_shrtct1.pdm (update mode, freeze version, and no extraction.) Select all actions.

6) In the Repository Browser, select Target model for either document

Note: Both remain at version 1 and they reference the orginal target model(s)

system.log ALL the Java System.properties.

This includes such information as the

java.class.path and the file.encoding (the

default character set). This may help when

troubleshooting.

properly checking for errors in the primary database.

Workaround: You will need to left mouse click on the Connection profile before the Profile Status is updated.

$SYBASE/shared-1_0/lib/ldap.jar

$SYBASE/shared-1_0/lib/dsparser.jar

$SYBASE/RMP-12_6/lib/asa.jar

$SYBASE/RMP-12_6/help/rmen.jar

$SYBASE/jConnect-5_5/classes/jconn2.jar

Note, $SYBASE should be replaced with the actual sybase directory.

text data that is null, while the original DROPPED replication definition has not null

defined for the text column. This will happen when mutiple repdefs are created on one table such that repserver creates an internal overall repdef.

Currently, RepServer requires that the sa login has the same password on the primary

database and replicate database when using 'create subscription' to materialize a table

subscription. RS logs in to primary as 'sa' and does a 'select' upon the table to

create a materialization queue of data.

( "." ) in it, the instance configuration file

is backed up to the incorrect directory.

//////////////

From the Customer

The tutorial area is the place to put the "why" and "how" information about this functionality. The user needs to clearly understand the distinctions between all of the repository functions (freezing, locking, branching, consolidating, extracting, etc). The current documentation is too focused on the "and then you click the button" step-by-step type descriptions, without enough context of "you would click the button if you wanted to...".

Digressing:

Overall the current tutorials for the entire suite are very weak. There have been previous cases logged about errors in the tutorial (when you follow the instructions for the tutorial, the tool doesn't respond as documented). I have submitted cases in the past about reworking the tutorial to be more of an interactive walk-through (308825). This would mean that the PowerDesigner interface would be "driven" by the tutorial, which would instruct the user to do certain things, and monitor what the user actually did, and either proceed (correct action) or clarify the step (incorrect action).

9/2/04

This issue has been addressed in PD 11.0.0.

To reproduce:

1) Create table with columns A, B, C (in this order)

2) Add a primary key with column order B, C, A

3) Consolidate/extract model

4) Change PK column order to A, B, C

5) Consolidate

Note: The move action is detected, but it is not selected. It should be.

9/20/04

This issue has been addressed in PowerDesigner 11.0.0.

Feature Request to allow rs_init to reference a specific interfaces file.

For Example:

rs_init -I <interfaces_file>

under Windows 2000. The code has been fixed,

but you may need to manually repair the .bat

scripts in %SYBASE%\rax-12_5\bin. For each

script, if you see a "%*" on the command line,

change it to enumerate the parameters instead:

%1 %2 %3 %4 %5 %6 %7 %8 %9

Also, to right-click on documents and have a menu item that will add the document to a configuration.

If all docs are V1 only, then just bring the list into the configuration. If any docs have multiple versions, bring up a list **of only the multi-version docs** in a dialog box, showing a "2 across grid" with the documents down the left side and their current version on the right. Each version cell field has a pull down of all valid versions of that document. Clicking OK will populate the member list as appropriate.

Customer's comments:

Basically, there needs to be improved functionality to be able to change which version of a specified document that participates within the members tab page of a configuration. The displayed list of documents could be displayed in a grid format or dropdownlist format, where the cell field holding the version number is editable (meaning this field lists all the version #'s in the current branch and the user can select which version they want in the configuration). Also, this feature would only be available to those Repository users/groups who have the proper permissions to create, delete, edit a configuration.

Tse/customer comments:

The Repository Right called "Manage Configuration" allows a user to create, modify, and delete configurations. However, the customer would like to have additional control over the "access permissions" to the configuration itself, consider the configuration as another object in the Repository like a folder or a document. Allowing the user who created the configuration or someone with more permissons (manage repository) to set the access permissions on a configuration...

This is a important feature to managing a team development environment... Let's say we have 20 documents in the Configuration, each document has a different version level. We are ready to release our application to production. The members of the Configuration at this given point (the production release time) contains all the versioned documents that represent the application version. We need to be able to know this collection equates to the production release and minimize anyone who has 'manage configuration' rights from changing it using the 'access permissons' to control that (list, read, write, full). Now, an update occurs to the production application. So, we have new versions of documents, so we want to change the version of the documents contained in the configuration. Today, we have to delete a document and then add it's newer version to the configuration. Now, our configuration is updated. The configuration is not a static thing... it becomes is a living/evolving archive of our production work. This is our thought process on the use of the configuration.

LTMLOG shows the following error:

> LTMMGR Received log message = I. Mon Dec 22 12:51:41 2003. 09093 - The setting for Codepage is '284' and ,force is 'being used'.

> LTMMGR Received log message = I. Mon Dec 22 12:51:41 2003. 09175 - The setting for RS_ccsid is '850'.

> LTMMGR Received log message = I. Mon Dec 22 12:51:41 2003. 09174 - Codepage and RS_ccsid combination not supported.

> LTMMGR Received log message = E. Mon Dec 22 12:51:41 2003. ERROR �12020 (H.R126.C(LTMMGR)) - //DSN:DSMITH.R126.C(LTMMGR)(332) 12020 - Configuration of the MVS LTM failed.

Jan 19, 2004

Ct tested successfully the ENG supplied pre-release of LTMCNTB link lib module with new codepage conversion from Spanish EBCDIC CP284 to RS Ccsid 850

+ DRL@LGGX fails in a "catch-all" trap in the branch table with the following entries

4780B56C 47F0B854 47F0B590 4700B590 00DEAD18

NOTE No DB2 utility is being used

As a result of an ongoing performance tuning exercise of the RAD installations, a customer has asked for additional information to be collected and displayed by the Task Monitor:

+ Wait time in mini-seconds per RAD component

+ The Name of the other Subtask being waited on, if applicable

property to "default" value.

To reproduce:

target DBMS: ASE 12.5

1) create a table with several columns

2) Add two triggers, ord_1 and ord_2, both using the delete trigger template

3) Ord_1's firing order should be set to 1. By default, the firing order for ord_2 is 2.

4) Consolidate and extract the model

5) Change the firing order for ord_2 to 1 and ord_1 to 2.

6) Consolidate with freeze current version

7) In the merge window, you must 'select all' actions ( not sure why, but the actions are not selected by default)

complete the consolidation

9) Compare version 1 and 2 of the model.

Note: they are identical.

Also, tested merging the models and the models merge successfully

1/22/04

This issue has been addressed and will be GA in the PD 10.1.0 maintenance release.

Chapt 13 - Managing Warm Standby Applications

Setting up warm standby databases

Both "dump database" and "dump transaction" can be used for dump marker

Change "...dump operation"

to "...dump operation: dump database or dump transaction"

...adding +1 to the real Maximum target length.

Reporting Maximum target length 11 when repdef column defined at 10.

creating and managing new error classes.

Workaround:

You will need to left mouse click on the RepConnector connection and then right mouse click on the RepConnector connection to display the correct menu list.

or warmstandby for DDL and subscription materialization.

Allow customers to specify 'login as' option to allow for different users/passwords.

7465869701.847564537284756) to DB2 DECIMAL values results in (1) values

that are over padded (e.g. 7465869701.8475645372847560000), (2) values

that are over truncated (e.g. 7465869701.8475645) or (3) the unexpected

termination of the Replication Server.

setting of it twice.

NOTE The LPAR must be re-IPL'ed to avoid the system 0C4 abend when restarting RAD.

Apr 19, 2004

SAS Eng requested sample SYSUDUMP

May 04, 2004

The Engineer in SAS, 3rd party vendor of SAS/C, tracked down the cause of the System 0C4 abend to C load module LTMVERSION being called by RAD's main LTMMGR program without having R1 initialized properly (i.e. in regular batch job scheduler's convention in passing a parameter area using R1). Sybase ENG will formulate correction and fix in due course.

StringBuffer to hold LTL commands before

sending them to RepServer. This has caused

problems with some binary and character data.

To correct this, RepAgent now uses a byte

buffer for LTL commands and does any required

charset conversion to *only* character data

instead of having jConnect do charset

conversion on the complete LTL command.

THIS FIX REQUIRES THAT YOU UPDATE YOUR

JCONNECT-5_5 TO EBF 11473 OR LATER.

Charset conversion is handled as follows:

1 - Be sure that the system default charset on

the machine on which RepAgent is running is the

same as that of the primary database. Sometimes

this is insufficient and you need to set it

within the RepAgent run script(s):

Edit the various RepAgent run scripts to add

the 'file.encoding' Java system property to the

Java command line. The property (including the

double-quote marks) should follow the memory

property and precede the 'install.root'

property. The format of the 'file.encoding'

property is:

"-Dfile.encoding=xxx"

where 'xxx' is the Java (not Sybase) name of

the default charset (note that the charset name

is case sensitive). For example, inserting the

following property in the Java command line in

the 'ra' script

"-Dfile.encoding=Cp1251"

sets the default charset for RepAgent to

Windows Cyrillic.

2 - If your RepServer is using a different

charset and you wish RepAgent to convert from

its default charset to the one used by

RepServer, set the 'rs_charset' property to the

same Sybase charset that the RepServer is

using. If you want RepAgent to send data to

RepServer using the default charset that you

have configured, do not set rs_charset.

Be aware that converting from one charset to

another is not recommended and may lead to data

corruption because of charset

incompatibilities.

Note that, for all non-Sybase primary

databases, setting 'pds_charset' does nothing

because those JDBC drivers do not provide a way

of setting the charset. Also, it is not

necessary to set 'rssd_charset' because the

jConnect JDBC driver will determine the charset

to use when the RepAgent connects to the RSSD.

This may affect database level replication.

----------------- lwp# 36 / thread# 36 --------------------

ff21b91c yield (1, 1, 16d018, 238264, 0, 0) + 8

00238274 srv_yield (0, 56e8e8, 56e9b8, 0, 8, 2000) + 14

0005d210 _sync_spinlock (29b3900, 447abc, 10, 29b3880, fe50e634, 1) + d0

0016dc68 _sqt_del_from_cache (29b36c8, 194ec30, 29ac308, 483084, 29b3478, 4ba828) + dc

00168fd0 sqt_delete_tran (29b36c8, 194ec30, 29ac308, 1, 1, 194ec48) + 40

00178690 _sqt_purge_open (29b36c8, 0, 29ac308, 25, 194ec30, 2f3bf70) + 38c

00168cd0 sqt_read_stmt (29b36c8, fe50fd28, 29ac308, 2000, 1, 29b3478) + 27c

000ad18c _dsi_read_queue (18cf838, 29b36c8, 404ca60, fe50fd28, 29ac308, 0) + 48

0009b6c4 dsi__ProcessSpecial (18cf838, 404ca60, 29ac308, fffbffff, 0, 20000) + 1a0

0009a1d4 dsi__DispatchGroups (18cf838, 29ac308, 0, 0, 20000, 1) + 58

00099720 dsi__SchedulerBody (18cf838, 29ac308, 0, 5, 3, 3b9c00) + 190

000994bc dsi__SchedulerWrapper (7, 3b9c00, 29b23e0, 554400, 29b23e0, 0) + 724

00237a44 srv__start_function (769770, 0, 0, 0, 0, 0) + 1c0

ff260694 _lwp_start (0, 0, 0, 0, 0, 0)

The stacktraces most often occur in either an SQM Writer thread or a Distributor thread and are typically related to the functions exc_unlink_ele(), _exc_alloc_element() or exc_link_ele().

After a crash upon reboot, Replication Server sometimes crashed immediately again, sometimes it will run for a while.

USER COMPLETION CODE=1204

LTMEXT attempted to obtain storage of length zero.

This happens when a dump command is in the queue and skipped by the DSI due to

default configuration and the next command brings the DSI down.

Workaround would by 2 times skip transaction

or poor performance.

Feature Request to limit rs_databases replication to global system changes,

create/drop/alter connection, and not replicate local RS information across route to other replicate RepServers.

"A parallel transaction has failed in database XX. The transaction will be executed individually."

4F530E43505002E6 01.

To reproduce:

1) Open project.pdm ( make note of the 'last modification date' on the Model Property version tab)

2) Consolidate and extract the model (the date is fine)

3) Wait 1 or 2 minutes and then add a new table to the model

(make note of the 'last modification date' on the Model Property version tab)

4) Consolidate and extract the model (the date is fine)

5) Extract the model to a new file or in memory

6) Compare the 'last modification date' of the new model with that of project.pdm

Note: The value will be that of the orginal original 'last modification date' of project .pdm.

to force valid segments onto another partition. This should allow partitions

stuck in ON-LINE/DROPPED/ state to be removed from repserver without having

to wait...

Customer Feature Request for...

configure replication server set disk affinity "partition_name"

when the disk is full.

SQLException "S0022 column not found" error.

resource message, two things happen:

- the quote mark does not show up in the

localized message

- if there is/are parameter placeholder(s) in

the resource message, then the actual values

do NOT show up in the localized message;

instead, the placeholders themselves show up

then the standby connection is dropped, the "admin logical_status" will report the

old standby is still available. The cleanup of the rs_database is not correct. The

situation can be resolved by updating the rs_database table to show a single row

with ltype = 'P'. The logical reference (ltype = 'L') can be removed.

(1) In RAD 12.6 Installation Guide: Chapter 3, Heading "Understanding DB2 database consideratios", Subheading "DB2 utilities"

>You may now use the DB2 LOAD LOG=YES utility to assist in replicate materialization.

(2) In RAD 12.6 User's and Troubleshooting Guide: Chapter 3, Heading "Task3", Subheading "Materialize replicate table", Topic "Using bulk utilization"

>Replication Agent for DB2 supports use of the DB2 LOAD LOG YES utility to assist in replicate materialization.

(3) In RAD 12.6 User's and Troubleshooting Guide: Chapter 4 "Replicating Source Table", subheading "Using bulk utilization", topic "DB2 Load Utility":

>Replication Agent supports use of the DB2 LOAD LOG YES utility, and it is one of the methods currently available for materialization.

to send tracing information to a seperate file and not the RS log.

"select convert (char(26), origin_time, 109) from rs_lastcommit..."

method to report latency time +/- 1/300th second.

additional indexes have been added to the

TranLog table and, where appropriate, queries

explicity use them. This fix changes the

schema and requires migrating the XLog.

MIGRATING the XLog:

If you are creating an XLog, the new schema

will be used.

If you have an existing XLog, you must do the

following to migrate the schema:

1 - login to the RepAgent and verify that

scripts run automatically:

ra_config pdb_auto_run_scripts, true

2 - shut down the RepAgent and apply this EBF

3 - start the RepAgent in ADMIN mode

4 - login to the RepAgent and issue the

pdb_xlog command with no parameters

When the command returns successfully,

migration will have been completed and you may

resume replication.

Note that, regardless of how many fixes in this

release require migration, you need migrate a

given RepAgent instance only once.

With this fix, Repserver will set SYBASE_REP to default value if it is not already set.

SYSMDUMP dump provides all desirable info but it is large (> 160Mb)

SYSUDUMP dump is a good compromise to meet most diagnostic requirements

Customer would like to be able to create a folder inside a specific branch and to let it visible only in the branch where it has been created.

create a model with some objects anme 'Axxx' and some others named 'axxx'

consolidate

search in repository A* or a* (case insensitive)

you'll find all objects

search in repository A* or a* - case sensitive

you'll find nothing

when LTI component stops.

For the same reason, the name of the log file is now based on the agent instance name (instead of system.log). For example: $SYBASE/MRA-12_6/my_instance/log/my_instance.log Also, the "log_trace_name" command is now hidden but still available for backwards compatibility; it returns the same value as the "log_system_name" command.

every few seconds, DSI stops, rollback, and individually executes

transactions every few seconds. This should not have happened; at the

default of 1000 DSI should wait 3 minutes not every few seconds.

This action resulted in very poor performance at the RDB DSI

the customer would like

A check box to select all or unselect all documents and a check box by document : to be able to extract what he wants like other selection dialog box inside PDesigner (example : Generate database --> Tab selection).

It could be also nice to have in this window, a check box by setting to be able to select the setting directly in the window extract multi documents instead to open per model its setting (example table --> properties --> Tab column)

because there is at least one operation listed

in the tran-log queue without a corresponding

entry in the shadow table data queue. When

RepAgent is re-started, replication works

just fine.

RepAgent now recognizes the mismatch between

the queues, prints an error message, and goes

to ADMIN state.

In addition, a new property called

'lr_attempt_recovery' is now available. The

default value is FALSE. However, if set to

TRUE, the RepAgent, upon recognizing the queue

mismatch, will attempt to retrieve the missing

shadow table data instead of going to ADMIN

state. Sybase recommends that you change this

property only on advice from Technical Support.

data if primary column is 'ntext'.

Since RepAgent 12.5 does not support the

RepServer 15.0 UNITEXT data type, when

replicating TO MsSQL, the RepAgent's encoding

of the character stream affects the byte order

of data when it is applied at the replicate.

For this reason, RepAgent now has a new

user-configurable property called

'lr_ntext_byte_order' that allows the user

to set the byte order that RepAgent should use

on ntext data -- either BigEndian or

LittleEndian.

For example, if the replicate is an ASE on

Unix, set the property to "big"; if the

replicate is a MsSQL on Windows, set the

property to "little". The default value of

'lr_ntext_byte_order' is "big".

during RS startup, RS may generate thread fatal

error messages regarding failures to acquire

locks due to RepAgent connection failures due to

time out. This behavior is acceptable. Once

queue processing is complete, RepAgents can

connect successfully. Therefore, remove the

generation of the thread fatal error messages

since they cause unnecessary concern to the

user.

and column names) that begin with a digit are

valid so should be delimited.

when that column is greater than 32K.

enough for DATE.

handle user synonyms of system datatypes.

Quiesce command should just suspend the RAT , It shouldn't kill RAT

Currently, ASE command "quiesce database" kills RepAgent

Scenario: version 9.5.2 PowerDesigner. was used to create repository tables that resides on Sybase ASE 12.5.x database (Unix platform), then the end user connects to the repository using version 10.x PowerDesigner to update it. The repository update script dialog pops up. The end user executes it and during the transactions, the generation log produces 2 errors on the last 2 UPDATE statements:

Errors:

----------

*** Executing statement 1032:

update PMEMDL set TOID=(select POID from PMOBJT where OBJT=TOID) where exists(select 1 from PMOBJT O where PMEMDL.TOID=O.OBJT)

[INTERSOLV][ODBC SQL Server driver][SQL Server]Invalid column name 'TOID'. SQLSTATE = S0022

*** Executing statement 1033:

update PMEOBJ set TOID=(select POID from PMOBJT where OBJT=TOID) where exists(select 1 from PMOBJT O where PMEOBJ.TOID=O.OBJT)

[INTERSOLV][ODBC SQL Server driver][SQL Server]Invalid column name 'TOID'. SQLSTATE = S0022

Resolution from PD Development:

-----------------------------------------------------

Run PowerDesigner's (auto-generated) update script outside of PowerDesigner UI (in an ISQL session) AND modify the 2 update scripts by adding in the alias (PMEMDL and PMEOBJ) for the TOID column in the first select statement as follows:

update PMEMDL set TOID=(select POID from PMOBJT where OBJT=PMEMDL.TOID) where

exists(select 1 from PMOBJT O where PMEMDL.TOID=O.OBJT)

go

update PMEOBJ set TOID=(select POID from PMOBJT where OBJT=PMEOBJ.TOID) where

exists(select 1 from PMOBJT O where PMEOBJ.TOID=O.OBJT)

go

9/9/04

This issue has been addressed in PowerDesigner 11.0.0.

message after being interrupted. For more

details, see description of fix to CR #365503.

result set headers and valid states in which

commands can be executed cannot be found

causing an error similar to the following when

the user tries to set properties or issue

commands:

Cannot load resource bundle,

com.sybase.ra.resource.RACLIResource,

for TDS header print out, <Can't find

resource for base name

com.sybase.ra.resource.RACLIResource, locale

iw_IL>

returns multiple rows when Oracle is parallel.

This is one of the doc bug.

Request for Enhancement per peck:

-----------------------

Extend "set autocorrection" semantic

to support autocorrection at a whole-database

level:

Currently, we only support autocorrection

for specifically-identified tables, as

follows:

set autocorrection { on | off }

for replication_definition

with replicate at data_server.database

Proposal:

Also Support:

set autocorrection { on | off }

for database replication definition db_repdefname

with primary at data_server.database

with replicate at data_server.database

The difficulty here is one of the various threads thrashing about in contention for an internal resource required to complete their initializations. This resolution addresses the issue by enforcing "tighter" thread serialization at startup and putting threads to sleep until it is their turn to initialize, thus reducing CPU requirements during this initialization phase.

Device Usage Map entry for database <4>, logical page id <-1> not found.

throw out un-replicated operations as "duplicates."

with same names but with different character case

(e.g., "MyColumn" and "mycolumn").

different owners.

procedure) to have more than one column (or

parameter) with the same name but different

character case (e.g., "myCol" and "MYcol"), then

when replicating, RepAgent replicates the value

of the first column to *both* columns.

illegal calls at the interrupt level if the trace OSRV, OS RV_TR_ATTN

is turned on. The trc_print functions that are called to write

messages require a mutex to be locked via a call to srv_lockmutex().

This is not allowed at the interrupt level by Open Server. This

results in an endless loop of attempting to report an error until

the server runs out of memory and crashes.

a. Browsing packages in the repository

b. Extract each package from it

c. for each of these models, create a report and generate html file

a table with all classes 'flatten', another with attributes,and a FK that refers to the classes tabels (instead the relationship table), ....

mra_admin utility to allow a Mirror Replication

Agent instance to be created and configured using

a resource file.

Resource files are text files that you can modify

with any text editor. They contain information

that identifies the resources and configuration

values to be used by the mra_admin utility to

create a Mirror Replication Agent instance.

To create a resource file, copy the template file

found in the init subdirectory of the Mirror

Replication Agent installation directory and edit

the copy as needed. The template resource file

contains descriptions of each of the values and

options that you can set.

Once the resource file is created, you can execute

mra_admin with the -vr or -r command line

parameter, specifying the resource file that

mra_admin will use for input.

New mra_admin parameters:

-r <resource> Creates a Mirror ReplicationAgent

instance based on the contents

of the resource file specified.

-vr <resource> Validates the contents of the

resource file specified, without

creating an instance.

Note: Sybase recommends that you validate a new

resource file by using the mra_admin -vr parameter

first, before you create the instance using the -r

parameter.

EXAMPLES:

USING RESOURCE FILES FOR MRA_ADMIN

This section explains how to prepare a resource

file for mra_admin.

1. Make a copy of the template file and rename

it to distinguish it from the original.

For example:

cp mra.rs pubs2.rs

2. Use a text editor to edit the new resource

file.

To use the vi editor, for example, enter:

vi pubs2.rs

See "Editing a resource file" below for

instructions.

3. When you finish editing the resource file,

properly set your Sybase environment

variables and start mra_admin from the

operating system prompt, using the -vr or

-r parameter. For example:

$SYBASE/MRA-12_6/bin/mra_admin -vr pubs2.rs

EDITING A RESOURCE FILE:

Resource files list the attributes to be defined

during an mra_admin session and the values for the

attributes. Attribute entries are in this format:

attribute=value

where:

attribute - is an item for which you are

supplying information, such as a

server name or network protocol.

value - is the value you want to assign to

the attribute.

mra_admin ignores blank lines and lines that begin

with a pound sign (#). No spaces should separate

the attribute, equals sign, or value.

If a Sybase-defined default value exists for an

attribute, you can specify the default with the

special value USE_DEFAULT.

RUNNING MRA_ADMIN IN VALIDATION MODE

Note: Before you can invoke Mirror Replication

Agent utility mra_admin, you must first make sure

that the Sybase environment variables are set.

Refer to the "Mirror Replication Agent

Administration Guide" section on "Using Mirror

Replication Agent utilities" for more information.

Each execution of mra_admin with the -vr (validate

resource file) parameter returns information about

the validation and the result. When complete, the

last message displayed is "Admin processing

completed" or "Validation encountered failures."

If you encounter the "Validation encountered

failures" message, review the previous messages to

determine the exact cause of failure, correct the

necessary values in the resource file, and re-run

the validation until it executes successfully.

You can re-run the validation process as many

times as necessary. No entities are changed or

created as a result of this process.

RUNNING MRA_ADMIN IN CREATE MODE

Each time you execute mra_admin with the -r

(resource file) parameter, it attempts to create a

Mirror Replication Agent instance using the

resource file as input.

Note: Executing mra_admin with the -r parameter

creates files and other resources. Any attempt to

re-run this command will fail unless you first

remove those resources as described at the end of

this section.

Each execution of mra_admin returns information

about the instance being created and the result.

When complete, the last message displayed is

"Admin processing completed" or "Admin processing

completed with errors". If the "Admin processing

completed with errors" message is returned,

review the previous messages to determine the

exact cause of failure, and correct the necessary

values in the resource file.

Note:Before you re-run the mra_admin with the -r

parameter, delete any instance created during the

previous execution by using the mra_admin -d

(delete)parameter:

$SYBASE/MRA-12_6/bin/mra_admin -d <INSTANCE_NAME>

Example:

$SYBASE/MRA-12_6/bin/mra_admin -d pubs2_mra

startup with the error "ERROR #7035 BROKER(1) - /mem/mem.c(242)

Additional allocation would exceed the memory_limit of '20971520'

specified in the configuration." Increasing the memory limit via the

'memory' configuration line in the RSMS configuration file has no

effect. This behavior is more frequently observed when RSMS invokes

multiple event scripts.

And/Or add additional indexes that will improve performance

of rsupgr.sql upgrade script

maintenance user unless it exists in sysusers.

The owner.table syntax gets generated incorrectly.

It should be owner."table" instead it generates "owner.table"

then during replication, a NullPointerException

may be thrown.

THREAD FATAL ERROR #60036 GLOBAL RS(GLOBAL RS) -xec/exec.c(365)

Failed to acquire lock on sync element 'rsinfo_global_rshandle_mutex'

a syntax error when the XLog create script is

executed.

whole byte buffer (including 0x00 bytes)

instead of only the LTL command within the

buffer.

Java max memory, all the executable scripts in

the $SYBASE/MRA-12_6/bin directory and the RUN

script for any instance created after applying

this EBF now contain a variable called

RA_JAVA_MAX_MEM. Before setting it to a default

value, the executable scripts first check to see

if it has already been set. This allows you to

customize it by setting the RA_JAVA_MAX_MEM

variable in either the instance RUN script

(MRA-12_6/<instance>/RUN_<instance>) or in your

environment.

Add delay between command_retry for deadlocks;

...add command_retry_delay parameter to DSI connection.

ODBC Driver : IBM ODBC driver or Datadirect wire protocol

When extracting a model from the repository, the following error message occurs:

SQL0104N An unexpected token "COUNT" was found following "".

Expected tokens may include: "+ ) - ". SQLSTATE=42601

This results in an unreadable file.

Request support to start repserver as a service

to "ignore", Errors 4985, 4986, 4987, 4988. Currently, these ASE errors are not

defined; defualt to "stop replication"

in permission failures during pdb_quiesce

processing.

> LEX0024 Connection established to DB2 subsystem DB2W.

> LEX0062 QID: E2E3D6D7000000000000000000000000000000000000000000000000130A1200

> LEX0025 Log processing to begin at log RBA D6D700000000.

> LEXTRACE Read past end of log, check the truncpt.

> LTM0015 LTM IS PROCESSING SHUTDOWN

> LEX0026 Sybase log Extract terminating with return code 0.

A:

+ The QID appears to contain the EBCDIC character string "STOP" (xE2E3D6D7)

+ All DB2 transactions pending replication before the unscheduled/forced LPAR re-IPL are not received at the Rep Server

incorrectly invoked the debug version instead

of the optimized version.

W. 2004/12/13 15:52:48. WARNING #13061 DSI(106 mcdon.fbidb) - seful/cm.c(5699)

Server mcdon is using 'utf8_nocase' sort order, which is different from this Replication Server's sort order.

repserver will generate default function string rs_select for it when we

drop a subscription with purge. However, if we create a customized function

string rs_select for it that does NOT match the subscription where clause,

then dropping the subscription with purge will cause DSI to shutdown with

the message:

"Cannot find a matching function string for function 'rs_select' and function string class 'xxx_class'. "

The correct behavior is for repserver to generate a default rs_select as

long there is no matching rs_select instead of shutting down DSI.

using a resource file when the Replication Server

physical and logical connections are the same

name.

INTERNAL ERROR #21 DIST(100119 ASEPRD01.SFINANCE_IT) - t/sqtext.c(1061)

Open server call to routine 'srv_putmsgq' failed.

I. 2004/12/22 01:02:15. Open Server error: Error: 16113, State: 0, Severity

10 -- 'Invalid object id found in srv_putmsgq'.

> Chapter 2 Installing Replication Server

> Heading Installing Replication Server

> Sub-heading Checking for a valid install

point 2 refers to an incorrect location for the vpd.properties file:

> Verify that the vpd.properties file has been written in the $SYBASE directory

should read

> Verify that the vpd.properties file has been written in the Windows directory (see enviroment variable "windir")

(2) In the product manual "Replication Server 12.6 Installation Guide for Windows" under:

> Chapter 2 Installing Replication Server

> Heading When you have finished installing

> Sub-heading Starting Replication Server and RSM Server automatically

the texts refers to incorrect locations where the RUN_RS<servername> and RUN_RSM<servername> batch files are generated by the installation process:

> c:\sybaseREP\REP-12_5\install\RUN_RSservername.bat

should read

> c:\sybaseREP\REP-12_6\servername\RUN_RSservername.bat

> c:\sybaseREP\RSM-12_5\install\RUN_RSMSservername.bat

should read

> c:\sybaseREP\RSM-12_6\servername\RUN_RSMSservername.bat

The resolution is to modify the repdef correctly.

has a password assigned to it. If the password is removed from the login in PD 10, the update

to PD 11 will succeed. This is the current workaround.

1/26/05

This issue has been fixed and made available in PD 11.0.0 EBF1.

third-party products/code is now installed

when the product is installed.

hold a lock on index pages and never move. These pages stays blocked

and phsical_io is stopped.

has changed. The QUIESCE command waits until all

data in the primary log has been read and sent to

Replication Server before moving the Replication

Agent to ADMIN state.

Previously, the command only waited until the

Replication Agent's internal queues were

processed before moving to ADMIN state.

Customers who need a more immediate transition

of the Replication Agent to ADMIN state should

use the "SUSPEND" command.

OS: Win 2000

Product: PD 10.1.0_1969

While consolidating the model, you will get the error:

Error: "Another object already exists in collection Parent Key"

The "as replicate column name" does not replicate to the RRS_RSSD with

the create route. RepDefs created later, after route is finished,

do have replicate column name in RRS_RSSD.

The following new configuration parm must be added for each member of a data sharing group where member is the DB2 member name and dsn is the dataset name of the BSDS for that member:

BSDS=member,dsn

In general this should be possible for all similar things, like DBMS and model resources.

property is no longer available.

In the Browser, the customer has a closed model. They then extract the same model from the repository with the 'merge document' and 'open document' setting selected. When the mdoel is extracted, the model is open, but its default diagram has not been opened. The customer would like this diagram opened.

will fail if the primary database is down.

the machine on which RepAgent is running is the

same as that of the primary database. Sometimes

this is insufficient and you need to set it

within the RepAgent run script(s):

Edit the various RepAgent run scripts to uncomment

the line(s) for RA_JAVA_DFLT_CHARSET and set the

value to the Java (not Sybase) name of

the default charset (note that the charset name

is case sensitive). The following examples set

the default charset for RepAgent to Cyrillic:

Windows (should be all on one line):

if %RA_JAVA_DFLT_CHARSET%. == .

set RA_JAVA_DFLT_CHARSET=Cp1251

Unix:

if [ $RA_JAVA_DFLT_CHARSET. = . ]

then

RA_JAVA_DFLT_CHARSET=ISO8859_5

export RA_JAVA_DFLT_CHARSET

echo " "

fi

truncation are inaccurate and numerous.

replication" is misspelled.

replication error occurs, the LogReader

SenderThread does not shut down properly,

causing LTL to be sent out of order to RepServer.

transactions need to be skipped.

This is a new feature request only:

Today, even setting Tools - General options-Repository-Merge Document to false, when customer extracts a model, the Merge Document is still available for checking. Customer would like this option to be disabled since a General Repository option (Merge Document) was changed.

Repserver may hang or crash due to memory corruption.

time a start replication is requested. From that

point on, the database-level repdefs are never

refreshed, so any new or altered database-level

repdefs are ignored until the Replication Agent

is stopped and re-started.

In RA.LINKLIB, rename LTMINFO to LTMINFO7. Rename LTMINFO6 to LTMINFO. In the PDRBIND job, change the member name from LTMINFO to LTMINFO6 and run the job.

Steps to reproduce

New Activity diagram

2 packages P1 P2

In P1 diagram D1

1 activity : A1

In P2 diagram D2

1 activity A2

decompose the activity

in the new diagram D3 add a new activity A3

Consolidate the model as ADMIN

In the repository : new User : U1

On P1 : permission add user U1 full permission copy permission to all children

on P2 : permission add U1 : read copy permission to all children

Disconnect ADMIN and connect U1

Extract the model

In P1 add a new activity A11

in P2 D2 add a new activity A22, in D3 (sub-diagram) add a new activity : A33

Consolidate with update

got a warning : which is correct

Merge windows, have a look

it is possible to add A11 which is correct

It is not possible to add A22 which is correct

But have a look at the diagram D2, it is possible to add A33 which is not correct

click OK

check the repository : A33 has been added in P2

Get following error consolidating to Ms Sql Server 2000 repository db:

[Microsoft][ODBC Driver][Microsoft Sql Server] syntax error converting the varchar value materialized to a column of data type integer. SQLSTATE = 22005.

Get the following error consolidating to ASA 9.0.2 repository db:

[Sybase][ODBC Driver][Adaptive Server Anywhere]Cannot convert MATERIALIZED to a int

SQLSTATE = 07006

note: it appears to be a datatype (int to varchar (254))change on the PMPDMTABL.otyp and PMPDMVIEW.otyp column between the v. 9.5.2 repository creation script and 11.1.0 repository update/creation script.

Unexpected error while executing the following action:

Changing value of attribute Source Text Offset of object

PD 11 and PD 11.1 ebf 2

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object

In all versions the error occurs at the same level

Delete Class "Stichprobenergebnis"

Delete Shortcut of Class "Stichprobenergebnis"

Repository : Oracle 8.1.7

ODBC driver used : Datadirect OEM 4.20 Oracle wire protocole, OEM 5., Oracle driver

Repro

\\neo\CIR\CIRMails\39xxxx\397618

ct is woring in 10.1 version but I also got the problem in 11.1 ebf 2 version

When doing the consolidation in 2 steps

- consolidation 1

Delete Class "Stichprobenergebnis"

Delete Shortcut of Class "Stichprobenergebnis"

both unchecked

-consolidation 2

Delete Class "Stichprobenergebnis"

Delete Shortcut of Class "Stichprobenergebnis"

checked

It is working

When consolidation with "create baseline version" it is working

Could not isolate much the problem

No ODBC Trace

Steps to reproduce

Consolidate the server.oom and extract it

Merge into server.oom the local.oom

Select all actions

consolidate with update the repository

in merge windows Select all actions

Have a look

Package 0_Diagramme --> Fachliches Objektmodell

classe : shortcut Stichprobenergebnis is going to be deleted

Package AM.pro Backend --> 0_Gesch鋐tsobjekte

Class : Stichprobenergebnis is going to be deleted

Error message

Unexpected error while executing the following action:

Changing value of attribute Creation Date of object

If both are unchecked no problem

PD 11

New PM informix 9

1 table : T1 with column C1

create one view (using Tool create view or not)

SQL Query

Select

case when T1.C1 = ' ' then 1 else 0 end case as C_view

from

T1

consolidate using datatdirectdriver without extraction

upgrade the rep to 11.1 and to 11.1 ebf 4 using Informix driver

extract using informix driver

Model graphic is empty

the objects have been extracted but the view has no columns anymore

When you add another column to the table and to the view (new column before and/or after the case expression: no problem anymore

But it is a strange behaviour because in the ct model, the views have many columns and I had the problem

you will find

\\neo\CIR\CIRMails\39xxxx\398220

my_model_1 1 table, 1 view with a case expression in SQL query : broken after extraction

FIKS_3_1_small_repro_33 small repro from ct's model : broken after extraction

FIKS_3_1_small_repro_33_45 which contains a views with only case expression (from ct model): OK after extraction

*** 26 August 2005 14:58:13 *** mtinevez ***

PD 11.1 ebf 2

Repository : Informix

Driver ODBC used : Informix ODBC driver

Steps to reproduce

PD 11

- consolidate the model which contains few tables, views and view references

- extract : result OK

- Upgrade to PD 11.1 and after to PD 11.1 ebf 2 using Informix ODBC driver

- extract the model consolidated with PD 11.

Have a look in the graphic, nothing, objects are there but they are not visible

view references parent and child : nothing

View column : no name

model used to do the test

\\neo\CIR\CIRMails\39xxxx\398220

so that they won't fail if the apps is running under another isolation level

enhanced to allow archived redo logs to be

accessed when required data no longer exists

in the on-line redo logs.

Configuring the Replication Agent to include

access to archive logs also removes the

requirement that Oracle automatic archiving be

disabled.

When the Replication Agent configuration is

changed to include archive logs, the Replication

Agent will no longer issue archive commands to

Oracle, requiring that automatic archiving

in Oracle be enabled, or that archiving be

performed by a mechanism other than

Replication Agent.

The following three new configuration options

can now be used to configure the Replication

Agent to include archive log usage, obtaining

the archive logs from the location specified

and optionally having the Replication Agent

remove the archive log files when they are no

longer needed for Replication.

1) pdb_include_archives

New configuration property pdb_include_archives,

when true, indicates that Replication Agent

should read Oracle archive log files in addition

to on-line redo log files. When false, only

on-line redo log files are read, and the

Replication Agent expects to control the Oracle

archiving process (Oracle automatic archiving

must be disabled).

Default value: false

Allowed values:

true: Enables reading of archived Oracle redo

log files from the path specified by

configuration parameter pdb_archive_path.

Configuration of Oracle automatic archiving is

supported under this mode. Removal of old

archive logs (no longer needed to support

replication) may be provided using configuration

property pdb_archive_remove.

false: Only on-line redo logs files will be read.

Oracle automatic archiving must be disabled.

The Replication Agent will execute Oracle archive

commands to archive the redo logs once they are

no longer needed for replication.

Comments: Set this configuration to true when

use of archive logs in addition to on-line redo

logs is preferred or when Oracle must be

configured to perform automatic archiving. Set

this value to false if accessing only the on-line

redo logs is preferred.

2) pdb_archive_path

New configuration property pdb_archive_path

identifies the directory path where the

Replication Agent will expect to find

archived Oracle redo log files. Valid only when

configuration parameter pdb_include_archives is

enabled.

Default value: none

Allowed values: A valid directory path on the

machine hosting the Replication Agent, pointing

to a location where Oracle will place archived

redo logs.

Comments: Setting of the configuration parameter

is required and must be set to a valid location

before the Replication Agent can be placed in a

replicating state when configuration parameter

pdb_include_archives is also set to true. If the

Replication Agent cannot find an expected log

record in the Oracle online redo logs, the

Replication Agent will search this directory for

the archived log file containing the required

record. Archived log files can be removed from

this location by the Replication Agent, if

configuration parameter pdb_archive_remove is set

to true.

WARNING: The Replication Agent process must have

access to this directory, and 'read' permissions

on the individual archive log files. Access

failures will prevent the Replication Agent from

maintaining a replicating state.

3) pdb_archive_remove

New configuration property pdb_archive_remove,

when true, indicates that Replication Agent

should remove archived log files from

the directory specified by pdb_archive_path

during log truncation processing. Valid only

when pdb_include_archives is enabled.

Default Value: false

Allowed Values:

true: Enables removal of archived Oracle redo

log files from the path specified by

pdb_archive_path. Removal occurs based on the

execution of command pdb_truncate_xlog, or the

timing of automatic truncation based on

parameters truncation_type and

truncation_interval

false: Disables Replication Agent removal of

archived Oracle redo log files.

Comments: Set this configuration to true when

the path specified by pdb_archive_path is

established solely for Replication support, and

automatic removal of unneeded archived log files

is desired. If the path specified

by pdb_archive_path is shared by other processes,

or the removal of archived log files is expected

to be performed by processes other than the

Replication Agent, this parameter should be

false. Parameters truncation_type and

truncation_interval, and command

pdb_truncate_xlog have no impact when this

configuration parameter is set to false.

WARNING: When the pdb_archive_remove parameter is

set to TRUE, the Replication Agent process must

have UPDATE access to this directory, and

'UPDATE' permissions on the individual archive

log files. Failure to provide the necessary

permissions will prevent the Replication Agent

from removing old archive files.

After the three new configuration options

have been properly set, Replication Agent

command ra_updatedevices should be executed

prior to resuming replication. This command will

query Oracle to obtain the 'log_archive_format'

value from the v$parameter table. This format is

required to allow the Replication Agent to

recognize the archive log file names.

Once ra_updatedevices command has been run and

the three new configuration parameters have been

set and permissions to the archive log files and

directory have been granted, the Replication

Agent can be placed in Replicating state

and automatic archiving in Oracle can be enabled.

execution.

(ra_config log_directory, xxx) where "xxx" is an

invalid directory name, no error is returned.

property to the default value, the resulting path

does not include the instance name.

Benefit:

A repository creator can customize or delete these more easily.

The repository setup scripts could contain a few lines of metadata, similar to a "generate database as script" generation option. Some explanations what to do next or in case of failure or incomplete setup could also be placed in the SQL comment.

In an effort to clarify this request the customer was assked the following questions:

What specifically are you looking for?

Other metadata such as

date created,

created by user,

a general description of the purpose of repository database,

suggestions for a naming convention,

security issues (database level),

size recommendations....

What do you want for 'what to do next'?

In case of a successful rep setup, Perhaps a hint for the username and password that must be used to connect the repository.

In case of an unsuccessful setup: Hints what to do besides the three

choices (ignore - ignore all - abort). Explanations what "ignore",

"ignore all" "abort" mean, respectively. What are alternatives (e.g.,

retry) and consequences

In what scenario are you encountering an incomplete repository setup?

In the general scenario. Clicking on Powerdesigner menus and dialog boxes.

Sometimes the rep setup script is 13000 lines long, sometimes it has

19000 lines (for no apparent reason)

Submitting the dialog box with the setup script to the server with the

repository database. Creates above mentioned cryptic error dialogs.

The repository setup process is pretty awkward. A *general* redesign

(perhaps by means of a wizard) would be appropriate, IMHO.

In order to use this new feature, the Rep Server and the Open Server/Open Client must be equal to or greater than version 15.0. Versions earlier than 15.0 still support these objects using the old maximum identifier length of only 30 bytes.

Customer's comments:

Current need:

Reporting of high level objects and information against the models in the Repository like, "How many CDMs, PDMs, OOMs are in the repository?", "What models are in which folder?" "What is the repository version number of each model?".

ENHANCEMENT:

----------------------------

A user could (with some effort) certainly derive the desired meta-data out of the existing sample models and documentation that PowerDesigner Development has to offer to produce queries. However, it would be more beneficial to the end users if you could provide a more simplistic form showing the mapping (linking) of these tables and columns to each other, or maybe a hierarchical document that shows how one could 'mine the data' of an object in a model type. This hierarchical document would show the descending levels of metadata. A white paper of sorts.

Additionally, when maintenance updates are provided and it affects the repository (new tables, new columns, new datatypes etc...) then provide what has changed in the 'What's New' documentation. (ie) Adding the dragging of models and folders feature to the repository browser was a big change but, no mention in the 'What's New' documentation of what changes were specifically made to which repository tables. As an end user, while I understand that structure changes need to be made to a repository structure to enable enhancement, I have a couple of concerns:

1. Is the MetaModel in the examples directory updated with the latest changes?

2. Will my reports (queries) still work and if not will I have to redo all of the initial discovery work with the model to determine how to modify my report (queries) to work again?

Please, give the end user a heads up (document changes) which can curb time and effort in addition to building trust in the tool set.

11/04/2005

This enhancement has been reviewed and will not implemented. Per engineering, the repository metamodel.pdm in Example directory must be used to understand repository structure.

(ra_config log_directory, xxx) where "xxx" is

either an empty string or a space, the RepAgent

will hang on startup.

the Replication Agent may have insufficient size

to accommodate a primary table column defined as

type number without explicit precision.

The fix provided in this ESD affects only newly

created shadow tables (created when an object

is marked for replication). For shadow tables

that already exist, the column size will only

be corrected if the object is un-marked and

re-marked. Note that un-marking and re-marking

should only be performed if all data from the

affected tables has been replicated.

To manually perform the modification, without

un-marking and re-marking, the size of the shadow

table columns may be increased by altering each

column so that the shadow table columns size is

increased to varchar2(40).

Example:

ALTER TABLE <my_tab> MODIFY (colx varchar2(40));

Note that this modification is necessary only if

the existing shadow table column size is less

than 40 for a primary table column defined as

NUMBER without any precision.

replication, you must set the configuration

property rs_charset to match the character set at

RepServer. If rs_charset does not match

RepServer's character set, then RepServer will do

character set conversion on the complete LTL

commands it receives which will corrupt the origin

QIDs and transaction IDs.

as warm standby.

> 290 SYSTEM COMPLETION CODE=0C4 REASON CODE=00000004

> 290 ACTIVE LOAD MODULE ADDRESS=0E800390 OFFSET=00040AFA

> 290 NAME=LTMMGR

This problem has been fixed.

Today, when deleting an object from a model directly in repository, customer would like a warning to be displayed telling that there are respective shortcuts in other models in repository that also need to be deleted.

In order to support inclusion of the time zone displacement, new configuration parameter pdb_timezone_file has been added to allow the customer to specify the exact location of the Oracle timezone file to be used.

This configuration parameter must be set to $ORACLE_HOME/oracore/zoneinto/timezone.dat or timezlrg.dat, depending on which timezone file the Oracle server has been configured to use.

and "log_append_existing" are supported in

exclusive debug logs.

When consolidating OOM model into repository for the 2nd time with "Freeze after consolidation" set, actor, message and objects are moved down improperly inside diagram. Other objects like rectangle are kept in correct place.

setting (the 'pdb_setrepddl' command) is reset

after invoking the 'pdb_xlog init, force' command.

The customer is requesting the same functionality for the result set of the Repository find objects.

"pdb_xlog init" fails resulting in the following error message in the log:

Work is being rolled back because of problems executing script:

ORA-01031: insufficient privileges.

fails with a BufferUnderflowException.

A model in repository has package1 and package2. Package1 has a shortcut dependency from Package2.

if we uncheck "Extract Dependencies" and extract Package, obligatorially PowerDesigner displays: "Your package selection has been automatically extended to include target packages referenced by internal shortcuts".

According to PD help, no shortcut dependencies should be included in package extraction.

different replication name with the command

"pdb_setreptable <priName>,<repName>,mark" or

"pdb_setreptable <priName>,<repName>,mark,owner",

replication fails because RepAgent does not use

the replicate name. This fix requires

Replication Agent migration.

ALL PRIMARY DATABASES: How you specify the owner

of the table in a replication definition has

changed from that described in the Reference

Manual. Now, you must always use the "owner"

keyword if you want to enable the SEND OWNER mode.

When marking, if you do not specify the replicate

owner, it will default to the primary owner.

In the following examples, the leading letter

indicates either primary (p) or replicate (r):

Example 1:

pdb_setreptable ptable, rtable, mark, owner

the table in the replication definition will be:

powner.rtable

Example 2:

pdb_setreptable ptable, rowner.rtable, mark,owner

the table in the replication definition will be:

rowner.rtable

Example 3:

pdb_setreptable ptable, rowner.rtable, mark

the table in the replication definition will

not be owner-qualified:

rtable

> 14.57.37 STC01191 LTMINFO Replication stopped for TWCLRFXD because data capture changes none

> 14.57.37 STC01191 LTMINFO: TABLE LOAD COMPLETE

...

> 14.57.38 STC01191 LEX0068 Replication updated for table JEDB21 .TWCLRFXD

> 14.57.38 STC01191 LEX0070 Column ERROR_CODE changed to 16

to clear the stored locator value.

removing replication definitions at RepServer.

This feature is included not only as a command,

but replication definitions can also be

automatically created when tables, or procedures

are marked, and removed when they are unmarked.

New command "rs_create_repdef" creates replication

definition at the configured RepServer.

Description:

Creates a replication definition at RepServer

for a marked table/procedure, or all marked

tables/procedures.

Syntax:

rs_create_repdef <ALL>

rs_create_repdef <TABLE_NAME>

Usage:

- When rs_create_repdef is invoked and the

parameter "all" or "ALL" is entered, a

replication definition is created for all

tables and procedures that are marked for

replication.

- When rs_create_repdef is invoked and the

name of a table or procedure that is marked

for replication is entered, a Replication

Definition is created for that table or

procedure.

- A result set is returned for each table

or procedure that a replication definition

create is attempted. The result set

contains the replication definition name

and status of the create. If the

replication definition was created, the

status will be "Created". If an error

occured, the error message from

RepServer will be returned for the status.

- The character case of the object names in

the replication definition will be set

according to the "ltl_character_case"

setting.

- Replication definition names for tables

always begin with the prefix "ra$", and end

with the suffix "_rd" and use the

replicate table name (possibly modified)

in between. All non-alphanumeric characters

except the underscore, and period are

removed from the replicate table name. The

period ("." ) is converted to an underscore,

and the resulting name is truncated, if

necessary, to 24 characters (repdef names

are limited to 30 characters). For example,

for a replicate name of "My Table", the

resulting repdef name is "ra$mytable_rd".

For an owner-qualified replicate name of

"wbeck.MyTable", the resulting repdef name

is "ra$wbeck_mytable_rd". For an

especially long replicate name of

"mytable89012345678901234567890"

(30 characters), the resulting repdef name

is "ra$mytable89012345678901234_rd".

- Replication definition names for procedures

are the same name as the procedure.

New command "rs_drop_repdef" drops a replication

definition at the configured RepServer.

Description:

Drops a replication definition at RepServer

for a table/procedure.

Syntax:

rs_drop_repdef <TABLE_NAME>

Usage:

- When rs_drop_repdef is invoked a replication

definition for that table is dropped at

replication server.

- A result set is returned for the table

or procedure that a replication definition

is dropped. The result set

contains the table name and status of the

drop. If the replication definition was

dropped, the status will be "Dropped". If

an error occured, the error message from

RepServer will be returned for the status.

- The character case of the object names in

the replication definition will be set

according to the "ltl_character_case"

setting.

- Replication definition names for tables

always begin with the prefix "ra$", and end

with the suffix "_rd" and use the

replicate table name (possibly modified)

in between. All non-alphanumeric characters

except the underscore, and period are

removed from the replicate table name. The

period ("." ) is converted to an underscore,

and the resulting name is truncated, if

necessary, to 24 characters (repdef names

are limited to 30 characters). For example,

for a replicate name of "My Table", the

resulting repdef name is "ra$mytable_rd".

For an owner-qualified replicate name of

"wbeck.MyTable", the resulting repdef name

is "ra$wbeck_mytable_rd". For an

especially long replicate name of

"mytable89012345678901234567890"

(30 characters), the resulting repdef name

is "ra$mytable89012345678901234_rd".

A new property "pdb_auto_create_repdefs" allows

replication definitions to be automatically

created and dropped at RepServer for tables and

procedures when they are marked and unmarked

for replication.

Description:

If set to true, when tables and procedures

are marked for replication, a replication

definition is automatically created at

RepServer for that table or procedure.

Syntax:

ra_config pdb_auto_create_repdefs,true

ra_config pdb_auto_create_repdefs,false

Usage:

- When pdb_setreptable is invoked to mark a

table or tables, a replication definition

will be created at RepServer for each table

that gets marked for replication if this

property is set to "true".

- When pdb_setrepproc is invoked to mark a

procedure or procedures, a replication

definition will be created at RepServer for

each procedure that gets marked for

replication if this property is set to

"true".

- When pdb_setreptable is invoked to unmark a

table or tables, the replication definition

will be dropped at RepServer for each table

that gets unmarked for replication if this

property is set to "true".

- When pdb_setrepproc is invoked to unmark a

procedure or procedures, a replication

definition will be dropped at RepServer for

each procedure that gets unmarked for

replication if this property is set to

"true".

- When pdb_xlog is initialized and table auto

marking is enabled, a replication definition

is created for each table that is marked for

replication if this property is set to

"true".

- The character case of the object names in the

replication definition will be set according

to the "ltl_character_case" setting.

- Replication definition names for tables

always begin with the prefix "ra$", and end

with the suffix "_rd" and use the

replicate table name (possibly modified)

in between. All non-alphanumeric characters

except the underscore, and period are

removed from the replicate table name. The

period ("." ) is converted to an underscore,

and the resulting name is truncated, if

necessary, to 24 characters (repdef names

are limited to 30 characters). For example,

for a replicate name of "My Table", the

resulting repdef name is "ra$mytable_rd".

For an owner-qualified replicate name of

"wbeck.MyTable", the resulting repdef name

is "ra$wbeck_mytable_rd". For an

especially long replicate name of

"mytable89012345678901234567890"

(30 characters), the resulting repdef name

is "ra$mytable89012345678901234_rd".

- Replication definition names for procedures

are the same name as the procedure.

command-size calculation are invalid. The invalid

LTI statistics are:

Total bytes sent

Avg LTL command size

Avg Bytes/second during transmission

Avg 16k LTM buffer utilization (%)

Avg LTL commands/buffer

RETURNING keyword are not successfully

marked for replication.

fails with a NullPointerException.

ALTER TABLE DDL operations containing constraint

changes. This fix requires that the RepAgent

PDS_USERNAME user has the SELECT privilege in

Oracle on the SYS.CON$ and SYS.CDEF$ tables.

This fix also requires that you re-initialize

your instance using "pdb_xlog init, force".

In the repository the customer would like the ability to update the permission of a 'child' with the parent's permission. However, they do not want to lose the 'child's' permission.

For example:

In the repository, create two groups

Add two folders. 1 - folder is the parent and the second folder is the child

Add group_2 to the child folder

Add group_1 to the parent folder

From the property sheet of the parent folder, if you click the 'Copy permissions to child folder' tool, the permissions of the parent folder are copied and replace the permissions of the child folder. (this is expected.)

However, the customer would like to have the parent permissions updated to the child folder but to have an option to preserve the divergenet permissions already associated with the child folder.

procedures that contain an underscore in the name

may result in the following error: Could not list

metadata for table <WZB_TS_TST> because: Too many

entries found in <or920tm1> for <WZB_TS_TST>.

Specify owner.

vector" received when an insert statement using a

SELECT INTO clause contains a NULL value.

a procedure" error occurs when DDL for a procedure

with a BOOLEAN argument is encountered.

the transaction context does not exist." occurs

too frequently.

operation with any of the system-defined types in

the USERDEF category like URITYPE, HTTPURITYPE,

ORDAUDIO, etc.

message received after altering an Oracle user

in the Primary database.

message received after altering an Oracle user

in the Primary database.

no option results in a ClassCastException after

applying Replication Agent/Oracle 12.6 ESD #1

Build 5200.

for a CREATE GLOBAL TEMPORARY TABLE command.

operation <> RBA <>: Failed to create a table.."

occurs for a table that is recreated.

Originally created 10/3/02 (PD 9.5.x?)and migrated PD 10, 10.1, 11.0 to PD 11.1.0

The customer is consolidating a PDM that contains views. Whether consolidating as a new model or as an update, the consolidation fails with the following ODBC error:

[S1010] [Microsoft][ODBC Driver Manager] Function sequence error (0)

[22018] [DataDirect] [ODBC Sybase Wire Protocol driver] [Sql Server] Implicit conversion from datatype 'INT' to 'VARCHAR' is not allowed. Use the CONVERT function to run this query.

Using the customer's repository, this issue is reproducable. Using a newly installed repository, the customer's PDM consolidates without problems.

In review of the ODBC trace file, the problem appears in the PMPDMVIWC table.

In the customer's repository, the 'CCMT' column has a datatype of 'Varchar(254)'.

In the newly created repository, this same column has a datatype of 'int'.

In both instances, the insert statement for this column is inserting an 'integer'.

09/30/2005

This issue has been addressed via CR 398220 which has been resolved in 11.1.0 EBF6

against CVData.java:81

OracleSessionContext sends RepAgent to

ADMIN state.

configuration setting (the 'pdb_gen_id' command)

is reset after invoking the 'pdb_xlog init,force'

command.

used to migrate the installed RepAgent version to

the current version.

Description:

Performs migration tasks (as necessary)

between releases of Replication Agent.

Syntax:

ra_migrate

Usage:

- When you apply an EBF of the Replication

Agent release, you may be required to run

this command to upgrade to the latest

version of the Replication Agent.

- When ra_migrate is invoked, it will migrate

the installed Replication Agent version to

the new current version.

- It will update the XLog system table with

the latest build version if the Replication

Agent was previously initialized.

- This command does not support migration to

an earlier version of Replication Agent.

- This command must be run in ADMIN mode.

a NullPointerException occurs.

to support localization.

However, sometimes the ASA charset has a slightly different name (especially on NT), if we can't find the same charset in ASA and the default for the OS is used.

In these cases, we print the following message:

W. 2005/12/08 16:28:08. WARNING #13092 GLOBAL RS(GLOBAL RS) - seful/cm.c(5833)

The Data Server 'erssd' is using the 'xxx' character set for

Database 'erssd'. The Replication Server is using the 'yyy' character set.

Unexpected results may occur.

Oracle database instance to locate the

timezone.dat file. This could be inaccurate under

certain conditions. Now, the agent first uses the

ORACLE_HOME environment variable, then the Oracle

database instance before reporting an error.

has filled up and that it is no longer being used.

Run rs_subcmp when Primary changes are not occurring.

"insert into select from with all NULLs "

is executed in Oracle, this can cause

Replication Server to fail with the error:

ERROR #1028 DSI EXEC(105(1) or920sh2.or920sh2) -

dsiqmint.c(3034)

command with a non-existent device id

results in a NullPointerException.

'pdb_automark_tables' now determines whether

or not the Agent automatically marks user

tables for replication. If the value is set

to true, user tables are marked during

initialization and DDL replication (when

replication of DDL commands is enabled).

If the value is set to false, user tables

are never automatically marked; they must

be explicitly marked using the

pdb_setreptable' command. The default

value is 'false'.

'dsi_sqt_max_cache_size' is set to a large value

(e.g. 1MB) and 'dsi_max_xacts_group' is set to a

value greater than 1.

file is not configured and ORACLE_HOME is not set,

RepAgent returns a garbled message: Replication

initialization failed because: Could not find

Resource Bundle containing index: FILE_NO_EXIST

and args=String index out of range: -1

NullPointerException.

"rs_zeroltm rao, pdb" is issued to

Replication Server system database and

"ra_locator zero" is issued in

Replication Agent, the Replication Agent

starts the log scan with an incorrect

locator.

stored procedures created with compile errors

return error code 4025. By default Replication

Server stops processing when this error code is

seen.

In order to allow Replication Server to continue

processing after receiving this error, Replication

Server must be configured to list a warning

message instead of stopping replication.

Issue the following commands to instruct

Replication Server to produce a warning message in

the Replication Server log when return code 4025

is received.

1 Create an error class for Oracle.

2 Initialize the new error class.

3 Assign the action 'warn' for error 4025.

4 Alter any Replication Server connection

that replicates TO Oracle to use this

new error class

1 Issue this command in Replication Server

to create the oracle error class

>create error class oracle_error_class

>go

2 Issue this command in the RSSD

to initialize the oracle error class

>rs_init_erroractions oracle_error_class,

>rs_sqlserver_error_class

>go

3 Issue this command in Replication Server

to update the oracle error class

>assign action warn for oracle_error_class

>to 4025

>go

4 Issue this command in Replication Server

to alter an existing connection to use

the oracle error class just created

after changing {pds}.{pdb} to the

appropriate values for your connection.

>alter connection to {pds}.{pdb}

>set error class to oracle_error_class

>go

procedure replication, Oracle Sequences can be

individually marked for replication using the new

Replication Agent command pdb_setrepseq.

pdb_setrepseq

Description: Returns the sequence replication

marking status; marks specified sequence for

replication; unmarks all marked sequences or a

specified sequence; enables or disables

replication for all marked sequences or a

specified sequence. This command is available for

Oracle only.

Syntax:

To return sequence replication marking status:

pdb_setrepseq

[sequence_name|mark|unmark|enable|disable]

To unmark, enable, or disable all marked

sequences:

pdb_setrepseq all, {unmark[, force]

|enable|disable}

To mark, unmark, enable, or disable a specified

sequence:

pdb_setrepseq sequence_name, {mark|unmark[, force]

|enable|disable}

To mark a specified sequence for replication with

a replicated name:

pdb_setrepseq sequence_name, repname, mark

Parameters:

sequence_name

The name of a user sequence in the

primary database. The sequence_name option can be

delimited with quote characters to specify the

character case. If mixed character case (both

uppercase and lowercase) is required, the name

must be delimited. For example: "Sequence".

The sequence_name parameter can be owner-qualified

to include the primary sequence owner name, with

each element separated by a period. For example:

owner.sequence

Note If you must use an object name case that does

not match the setting of the ltl_character_case

parameter, the object name must be delimited. If

an object name contains any non-alphanumeric

characters, such as spaces, periods, and so forth,

it must be delimited with quote characters. For

example:

"sequence name"

owner."sequence name"

repname

The replicated name of the sequence to be updated

at the replicate site, if desired to be different

than the sequence name at the primary site. The

repname option can be delimited with quote

characters to specify character case. See the

previous description of the sequence_name

parameter for details. By specifying a replicated

name, sequence updates can be replicated to a

sequence in the replicate database that has a

different sequence name from the primary database.

The repname option can be owner-qualified to

include the replicate sequence owner name, with

each element separated by a period. For example:

repowner.repname

all

A keyword that refers to all user sequences in the

primary database. By using the all keyword, you

can unmark all user sequences, or apply an enable

or disable operation to all marked sequences.

mark

A keyword that refers to marking user sequences

for replication.

unmark

A keyword that refers to unmarking marked

sequences.

force

A keyword that refers to the unmark operation.

When the force keyword follows the unmark keyword,

the pdb_setrepseq command immediately unmarks the

specified sequence in the primary database,

without first checking the enable status of the

sequence. When the force keyword follows the

unmark keyword and the all keyword, the

pdb_setrepseq command immediately removes

replication marking from all marked sequences in

the primary database, regardless of their enable

status.

enable

A keyword that refers to enabling replication for

marked sequences.

disable

A keyword that refers to disabling replication for

marked sequences.

----------------------------------

Understanding Sequence Replication

Logging of sequence information

Individual sequence changes are not logged in the

Oracle database log file. But changes to Oracle

Sequence do impact (update) the Oracle sys.seq$

table. These changes do not occur with each new

sequence value generated. Instead, the sys.seq$

table is updated periodically, based on sequence

caching refresh activity or other system changes.

The value stored in the sys.seq$ table for a

sequence is the 'next' value to be assigned

'after' the existing cache of values has been

exhausted.

As an example, a newly created sequence starts

with a value of 1, increments by 1 and has a cache

value of 20. These are all default values and can

all be customized. The value stored in the

sys.seq$ record for this new sequence is 21.

This indicates that the 'next' value to be used by

the sequence, after the existing cache of 20

numbers is used, is 21. The record in sys.seq$

does not change until the sequence value hits 21.

At that time, Oracle will 'cache' the next 20

values for the sequence, and the sys.seq$ record

will be updated to 41. It is this value (41)

recorded in change to the sequences sys.seq$

record, that will be used for replication. The key

point then is recognizing that not every

individual sequence update is recorded in the log,

and is therefore not available for replication.

Replicating sequence changes

When a sequence is marked for replication, changes

to that sequence against sys.seq$ are captured and

sent to Replication Server in the form of

parameters passed to a procedure. The procedure

(rs_update_sequence) must be installed at the

standby site as part of system setup, as well as a

function replication definition for that

procedure. At the standby site, an implementation

of rs_update_sequence will increment a same-named

sequence until its value is equal to that at the

primary.

Scripts are provided to create the

rs_update_sequence stored procedure and function

replication definition. The scripts can be found

in the Replication Agent installation as:

$SYBASE/RAX-12_6/scripts/

oracle_create_replicate_sequence_proc.sql

$SYBASE/RAX-12_6/scripts/

oracle_create_rs_sequence_repdef.sql

Performance considerations

Compared to the performance of incrementing a

sequence at the Primary database, particularly

where sequence values are cached, the effort to

increment the same sequence at the standby site

may be less efficient. The stored procedure must

dynamically determine the sequence to increment,

and must loop internally, incrementing the

sequence until the primary value has been reached.

The loop is required because there is no way to

'assign' a specific value to a sequence.

Since the name of the sequence is passed as a

parameter, Oracle cannot pre-compile the procedure

for efficiency. With the addition of the looping

activity required to properly increment the

sequence, the performance of the solution may

impact some environments where a large number of

highly used sequences is the norm.

Sequence replication alternatives

If the performance of sequence replication is a

concern, there are other alternatives to

replication that support primary and standby use

of the same sequence. These alternatives are

currently suggested by Oracle and others

interested in providing sequence coordination

between multiple sites.

1) Assuming the sequence is being used to generate

primary key values, the sequence at each site can

be concatenated with something unique to the site.

For example, use a sequence number concatenated

with the database name, site name or something

similar. This technique allows each site to

maintain a unique 'range' of sequence of numbers.

Each site having a unique range would mean there

would be no value in sending (replicating) changes

of one site's range to another site.

2) Similar to concatenating, each site could

obtain a different range of numbers by having

different starting points, or increment values,

for the same sequence. As an example, the sequence

at one site could start at one and increment by

two to generate odd numbers (1, 3, 5, etc) while

the other site starts at two and generates even

numbers (2, 4, 6 etc). Again, each site would

have a unique range and could avoid any need for

replication.

3) A third option is available to standby

solutions, where the standby site is for read-

only, and does not access the sequence value until

fail-over. Rather than continually replicating a

sequence's value, the value of the sequence at the

standby site could be updated as part of the

failover tasks. After failover and before the

standby allows connection to client applications,

a script or procedure could query the last used

sequence value (based on the last table to use it

for a primary key) and update or redefine the

sequence once, based on that calculated value.

a non-numeric value for the version or

field id field results in a

NumberFormatException.

hds_oracle_setup_for_replicate.sql is missing

synonyms required for DDL replication.

replicate when pdb_convert_datetime is false.

out the repository if the connection to the

primary database is not available.

error ORA-01000: maximum open cursors exceeded.

If redo log truncation fails, Oracle will

eventually run out of log space, which will cause

all primary database activity to stop.

OCS deferred queue size.

To close the open/orphan connections, either shutdown the PD application or kill the processes in Sql Server.

'eneric/useful/sync.c' with debug binary.

'truncation_type' is incorrectly initialized to

'interval', not 'locator_update'.

a delimited name fails with an error similar to:

Table <"qa1_Char"> could not be marked because:

Table <"qa1_Char"> is not in database <or920sh3>.

command does not update the devices

information.

DDL operation is unnecessarily filling up the log.

always using iso_1 for char/varchar column

encoding, regardless of the default JAVA

file.encoding or the RA_JAVA_DFLT_CHARSET

environment variable setting.

followed by re-marking of the same table

allowed un-marked rows to be replicated.

Oracle 9.2.0.6 UTF8 with the 9.2.0.5 and 9.2.0.6

Oracle JDBC drivers. The bug occurs when selecting

from a column defined as char(1) -- instead of the

select returning a single character, it returns

the character plus a blank (for example, "A ").

The workaround is to use the Oracle 10.1.0.2

JDBC driver.

This could be confusing to the user who would expect to see all of the "feature" languages (which are listed) as well as the "component" languages which are not.

followed by re-marking of the same table

failed to reflect change in owner mode.

Replication Agent as a background process in a

Korn or Bourne login shell, the Replication Agent

System Database (RASD) would shut down when

logging off or causing other signals to be raised.

timezone.dat file is reported after the Oracle

database metadata is loaded.

data types can not be marked for replication.

The unsupported data types for tables are:

SYS.UriType, SYS.DBUriType, SYS.HttpUriType,

SYS.FTPUriType, ORDSYS.ORDAudio, ORDSYS.ORDImage,

ORDSYS.ORDVideo, ORDSYS.ORDDOC,

MDSYS.SDO_GEOMETRY, all reference types, all

opaque types, the nested table and VARRAY

collection types and the MLSLABEL type.

The unsupported data types for procedures are:

all user-defined data types, including object

types, all reference types, all opaque types,

the nested table and VARRAY collection types,

the MLSLABEL type and the BOOLEAN type.

In order to do this:

Go to %SYBASE%\%SYBASE_SYSAM%\bin

and run

sysam uninstall

sysam install

This will install and start SYSAM service using Replication Server 15.0 release.

does not return all marked tables.

the owner filter list and the RepAgent is

restarted, the objects owned by the user

are not filtered out for replication.

be incorrectly read. Fix requires that

pds_username have 'SELECT' privilege on table

SYS.COLTYPE$.

Agent instance in the background causes the

Replication Agent System Database (RASD) to

shut down when logging off or causing

other signals to be raised.

with a replicate name, the replicate name is not

being correctly set.

marked for replication, the transactions

prior to the marking are also getting

replicated.

Oracle ROWID was missing.

object name in the error message is that of the

primary object, not the replicate object.

Oracle are incorrectly formatted.

fails in Replication Agent.

9i versions.

specified as the PDS user in the pds_username

configuration parameter is on the owner filter

list.

because of TXCTXT_LOB_PROCESSING_TWO_LONGS error,

which means the RepAgent thinks it has found a

table with multiple long or long raw columns.

when marking while in replicating state.

replicating UDD with numeric attribute.

when a procedure is dropped and recreated.

marked for replication is recreated

with the wrong owner.

replication, and subsequently deleted in the

primary database, DDL processing will remove

the marked article, but the shadow table is not

removed.

replication.

"Could not find Resource Bundle containing index:

ORAHOME_NO_EXIST".

12:37:14 (SYBASE) Vendor daemon can't talk to lmgrd (Cannot read data from license server (-16,287:22 "Invalid argument"))

12:37:14 (lmgrd) Please correct problem and restart daemons

To correct this, edit $SYBASE/SYSAM-1_0/bin/startd.sh as follows:

#!/bin/sh

# startd.sh

# P1 = component directory

ulimit -n 1024

ulimit -H -n 1024

cd $1

$1/bin/lmgrd -c $1/licenses/license.dat -l $1/log/lmgrd.log \

2> $1/log/stderr.out &

Then restart the daemon using this command:

$SYBASE/$SYBASE_SYSAM/bin/startd.sh $SYBASE/$SYBASE_SYSAM

the default owner filter list in parameter

owner_filter_list.

processing Sequence change on Oracle 9.0.1.

processing Large LOB.

stored procedure parameter type of ROWID.

DDL for new procedure from being processed.

has changed. The QUIESCE command waits until all

data in the primary log has been read and sent to

Replication Server before moving the Replication

Agent to ADMIN state.

Previously, the command only waited until the

Replication Agent's internal queues were

processed before moving to ADMIN state.

Customers who need a more immediate transition

of the Replication Agent to ADMIN state should

use the "SUSPEND" command.

'hds_oracle_funcstrings.sql' is not ERSSD

compatible. If you are using Replication Server

12.6 ESD #6 or earlier, then use the script that

is in the RepAgent installation scripts directory

(after applying this EBF). If you are using a

later release of Replication Server, then use the

Replication Server version of this script.

command fails with a Null Pointer Exception

command fails with error "Article

with id <> not found".

"The position <> is out of bound for the article <>"

is seen when processing a table row that is very wide

(has many columns and/or many columns populated

with large values).

fails with ClassCastException.

with more than one unique index fails

in Replication Server because the 'where'

clause contains no value.

This problem is related to Oracle behavior,

and what supplemental data is written to the

Oracle redo logs.

If a table has no primary key column defined, and

that table has more than one unique constraint

defined, Oracle will record only one of the

constraint column values in the redo log. The

selection of which column will be logged is

determined arbitrarily by Oracle.

To insure that a table with no primary key

column has the correct primary key designation

in any replication definition for that table,

the replication definition primary key list

should include ALL of the columns represented

by unique constraints. This will allow

Replication Server to use whichever constraint

columns are logged for that table as the

primary key.

build number in the repository.

supports Oracle replication (defined to

rs_username configuration parameter) requires

"CREATE OBJECT" permission be granted in

Replication Server. This permission is

required in order to create replication

definitions automatically or using the

rs_create_repdef command.

TransactionException: User with id <2> not found.

is seen. The stack trace includes the

parseSupplementalData signature.

2006-03-10 12:57:16,520 INFO [main] RMIService(136) - Starting RMI Service...

2006-03-10 12:57:16,629 ERROR [RMIRegistryThread] RMIRegistryServiceStarter(34) - Failed to start rmi registry.

java.rmi.server.ExportException: Port already in use: 9999; nested exception is:

java.net.BindException: Address already in use

at sun.rmi.transport.tcp.TCPTransport.listen(Unknown Source)

at sun.rmi.transport.tcp.TCPTransport.exportObject(Unknown Source)

at sun.rmi.transport.tcp.TCPEndpoint.exportObject(Unknown Source)

at sun.rmi.transport.LiveRef.exportObject(Unknown Source)

at sun.rmi.server.UnicastServerRef.exportObject(Unknown Source)

at sun.rmi.registry.RegistryImpl.setup(Unknown Source)

at sun.rmi.registry.RegistryImpl.<init>(Unknown Source)

at java.rmi.registry.LocateRegistry.createRegistry(Unknown Source)

at com.sybase.ua.services.rmi.RMIRegistryServiceStarter.run(RMIRegistryServiceStarter.java:28)

Caused by: java.net.BindException: Address already in use

at java.net.PlainSocketImpl.socketBind(Native Method)

at java.net.PlainSocketImpl.bind(Unknown Source)

at java.net.ServerSocket.bind(Unknown Source)

at java.net.ServerSocket.<init>(Unknown Source)

at java.net.ServerSocket.<init>(Unknown Source)

at sun.rmi.transport.proxy.RMIDirectSocketFactory.createServerSocket(Unknown Source)

at sun.rmi.transport.proxy.RMIMasterSocketFactory.createServerSocket(Unknown Source)

at sun.rmi.transport.tcp.TCPEndpoint.newServerSocket(Unknown Source)

... 9 more

2006-03-10 12:57:17,799 ERROR [main] RMIService(142) - Failed to start RMI Connector server.

java.io.IOException: Cannot bind to URL [rmi://abonner-sun:9999/agent]: javax.naming.NameAlready

BoundException: agent [Root exception is java.rmi.AlreadyBoundException: agent]

with a length of more than 32K fail

with an "Invalid token length encountered"

error message.

fails to include the origin time.

definitions are not automatically dropped

when you unmark all the tables.

generated for a Replication Definition could

duplicate a name that was previously generated.

This duplication caused the Replication Definition

creation request for a subsequent object to be

ignored.

Replication definition names were generated as

follows:

Table name Replication Definition Name

--------------------------------------------

order$req ra$orderreq_rd

orderreq ra$orderreq_rd

my.orderreq ra$orderreq_rd

your.orderreq ra$orderreq_rd

Now, a unique identifier is added to the

replication definition name. The unique

identifier is a maximum of 8 characters,

and the total name length is a maximum of

30 characters.

Replication definition names are now generated as

follows:

Table name Replication Definition Name

--------------------------------------------

order$req ra$0x794a_orderreq

orderreq ra$0x7952_orderreq

my.orderreq ra$0xa154_orderreq

your.orderreq ra$0xa155_orderreq

table for replication that was created by

Replication Server.

the dump database command... not the bad transaction. The Workaround is to resume the connection

with a skip transaction to skip the dump database command, then run sysadmin log_first_tran to

log the correct transaction in the exceptions log.

<SessionNumber=?, SerialNumber=?, User=null>

to repository" is encountered.

was created incorrectly for a table

containing a column of type OBJECT.

not configured for "ddl in tran" succeeds of the source database

is a "logical connection," but fails with ASE error 2762 if the

source database is a "physical connection."

RepAgent to ADMIN state.

could not be successfully created for

a table containing a column of type

LONG RAW.

allow the use of Veritas raw partitions.

fails.

com.sybase.ds.ase.device.Row.<init>

while attempting to retrieve a row on a page

during replication.

ds.ase.device.Page.getBufferSlice() while

attempting to retrieve a row on a page

during replication.

exec_sqm_write_request_limit: default: 1MB, upper limit: 2GB

md_sqm_write_request_limit: default: 1MB, upper limit: 2GB

sub_sqm_write_request_limit: default: 1MB, upper limit: 2GB

rsi_batch_size: upper limit: 128MB

For exististing replication environments the user will need to manually change the above parameters detault values.

holding the RSSD database

white spaces on column data containing

only white spaces or leading/trailing

white spaces when it generates the

XML document.

on the size of a string literal, when

replicating events from messaging to

database, RepConnector is unable to

process LOB data containing more

than 4000 characters.

A "ORA-01704:string literal too long"

error will be returned by Oracle when

processing data greater than

4000 characters.

RepConnector is unable to parse Oracle formatted datetime.

background process in a Korn or Bourne login shell,

the Replication Agent System Database (RASD) would

shut down when logging off or causing other signals

to be raised.

Device <log_raw1> is not open" is seen after

initialization or re-initialization of the

Replication Agent.

command. When used with the ra_locator command,

the 'zero' keyword instructs the Replication

Agent to set the value of the LTM Locator stored

in the Replication Agent System Database to zero.

If the value of the LTM Locator returned from the

primary Replication Server is zero, the

Replication Agent uses the LTM Locator value

stored in the Replication Agent System Database

to determine the point in the transaction log

from which to start reading data.

If the value of the LTM Locator in the

Replication Agent System Database is also zero,

the Replication Agent starts reading the

transaction log based on the database's secondary

truncation point setting.

operations that are chained across multiple log

records.

When versions of ASE_DRECOVERY and RepServer differ, RepServer cannot use entered license.

com.sybase.ra.lti.formatter.ltl

.LTLFormatter.compareFields()

re-created was not re-marked for replication

properly.

command may return an error similar to the

following: Msg 32000, Level 20, State 0:

Server 'raevdevl', Procedure 'pdb_xlog init',

Line 1: Command <pdb_xlog> failed - String index

out of range: -3

"Could not set article <SYS.IND$>"

encountered during replication.

if SMP and parallel DSI are enabled.

is seen along with a "NoSuchElementException"

vector" occurred, including an

"IndexOutOfBoundsException".

Unknown replication status -1" error when a column

from the primary database did not match the

position of the Replication Definition for the

same column.

migrating from any version prior to release

12.6.4049P7 and starting replication, a Runtime

exception during date conversion will send MRA

to ADMIN state if the LRTRACE trace flag is

enabled.

To work around this issue, disable the LRTRACE

trace flag and re-start replication. Once the

truncation point is updated by Replication

Server, the LRTRACE trace flag can be re-enabled.

against table SYS.IND$

an error in Replication Server:

ORA-01401: inserted value too large for column

verifiable Replication Agent instances.

workarounds...

create a simple subscription to a simple repdef

...or...

create a subscription to a function replication definition

alter connection to data_server.database set

dist_sqt_max_cache_size to 'value'

After restarting the Repserver, you can check the newly configured value using the following command(verified with the field named 'SqtMaxCache'):

admin who, dist

available in Adaptive Server Enterprise 12.5.4.

When working in group and using Repository using models with replicated objects, the replicated objects lose the reference from the original object when source model objects to be replicated are changed, consolidated again, and then the target models are extracted.

from multi-byte databases, are

replicated as type 'image' to

Replication Server version 12.6.

"A commit is seen, but a multi-record

operation reference is still open".

with NullPointer Exception

supports Windows authentication only with the

Microsoft SQL Server 2005 JDBC driver version

1.1 (June 2006). For details on how to configure

RepAgent/MsSQL to use this driver, please see

the description of bugfix #438563.

The customer is upgrading from PD 11.0.0 <no ebfs> to 12.0.0.

During the upgrade, they get the following error message:

Msg 207, Level 16, State 4 Server 'SYBDEV32', Line 1 Invalid column name 'OTYP'.

In testing, this issue has been reproduced when the repository is installed on ASE.

For my tests, I used ASE 12.5.4.

To reproduce:

Repository DBMS = ASE 12.5.4

1) Create a repository on PD 10.1.0 EBF12

2) Updated the repository to 11.0.0 GA

3) Updated the repository to 12.1.0 GA

4) In the generation log, search for ODBC, it will bring you to the error.

*** Executing statement 2429:

update PMPDMVIEW set TTYP=(select pt.TDAT from PMTEXT pt where pt.TXID=OTYP)

[DataDirect][ODBC Sybase Wire Protocol driver][SQL Server]Invalid column name 'OTYP'.

SQLSTATE = S0022

the Microsoft SQL Server 2005 JDBC driver version

1.1 (June 2006) along with the DataDirect SQL

Server JDBC driver that is distributed with

RepAgent.

NOTE that in order to replicate from Microsoft

SQL Server 2005 requires that you use the

Microsoft SQL Server JDBC driver. To replicate

from Microsoft SQL Server 2000, Sybase recommends

that you also use the Microsoft SQL Server JDBC

driver instead of the DataDirect SQL Server JDBC

driver that is distributed with RepAgent.

If you wish to continue to use the DataDirect SQL

Server JDBC driver, be sure that the RepAgent

property pds_connection_type is set to "MSMERJDBC"

(without the quotes). For details, see the

Replication Agent documentation.

By default, new RepAgent/MsSQL instances that you

create after applying this EBF are configured to

use the Microsoft SQL Server JDBC driver. You can

use this driver on both Windows and Unix systems.

To use the Microsoft SQL Server JDBC driver:

--------------------------------------------

1. Obtain the Microsoft SQL Server JDBC driver

from here (the URL should be all on one line):

http://www.microsoft.com/downloads/details.aspx?

familyid=f914793a-6fb4-475f-9537-b8fcb776befd&

displaylang=en

Install the JDBC driver on the machine on which

RepAgent is running or where RepAgent can

access it.

2. Before starting the RepAgent, add the location

of the JDBC driver to the CLASSPATH environment

variable:

On Unix:

setenv CLASSPATH /path_name/sqljdbc.jar:$CLASSPATH

where path_name is the path where you installed

the Microsoft SQL Server JDBC driver.

On Windows 2000 or 2003, go to

Start-->Settings-->Control Panel-->System-->

Environment, and add the following to the

existing CLASSPATH environment variable, using

the semicolon (;) as the path separator, or

create the path in the User Variables panel:

drive:\path_name\sqljdbc.jar

where:

drive is the drive letter, and

path_name is the name of the path where you

installed the Microsoft SQL Server JDBC driver

Click Apply, then OK.

3. For an existing RepAgent/MsSQL instance to pick

up the new value of the CLASSPATH environment:

On Unix, stop the RepAgent instance, define the

CLASSPATH as described in step #2, then restart

the instance.

On Windows, stop the RepAgent, close the DOS

window, open a new DOS window, then restart the

RepAgent instance.

4. For an existing RepAgent/MsSQL instance, in

ADMIN mode, configure the pds_connection_type

to use the Microsoft SQL Server JDBC driver:

ra_config pds_connection_type, MSSQLJDBC

then stop and restart the existing instance.

For new instances of RepAgent/MsSQL that you

create after applying this EBF, MSSQLJDBC is

the default value.

5. In RepAgent, configure the following properties

(for details, see the Replication Agent

documentation) using values appropriate for the

primary Microsoft SQL Server:

ra_config pds_server_name, <server>

ra_config pds_port_number, <port>

ra_config pds_database_name, <database>

ra_config pds_username, <ra_user>

ra_config pds_password, <ra_pwd>

To use Windows Authentication with the Microsoft

SQL Server JDBC driver:

------------------------------------------------

When running RepAgent/MsSQL on a Windows platform,

you have the option of configuring RepAgent/MsSQL

to connect to Microsoft SQL Server so that it will

use Windows credentials to authenticate the user.

NOTE that this feature is available only when

RepAgent/MsSQL is running on Windows 2000 or

Windows 2003 and requires Microsoft SQL Server

2000 SP 3 (or later) or Microsoft SQL Server 2005.

To use Windows Authentication:

1. In your primary Microsoft SQL Server, add the

<ra_user> as a Windows-authenticated user,

including the user's domain as appropriate.

For details, please refer your Microsoft SQL

Server documentation. Be sure to also add the

<ra_user> to the primary database and grant

appropriate permissions.

2. On the machine on which the RepAgent/MsSQL will

be running, add <domain>\<ra_user> to the

Windows user account. If no domain, then add

just <ra_user> to the Windows user account.

3. On the machine on which the RepAgent/MsSQL will

be running, copy the sqljdbc_auth.dll file from

the Microsoft SQL Server JDBC driver location

to a directory on the Windows system path.

When you installed the Microsoft SQL Server

JDBC driver, the sqljdbc_auth.dll files were

installed in the following location:

<install_dir>\sqljdbc_<version>\<language>\auth\

NOTE: On a 32-bit processor, use the

sqljdbc_auth.dll file in the x86 folder. On a

64-bit processor, use the sqljdbc_auth.dll file

in the x64 folder.

4. On the machine on which RepAgent/MsSQL will be

running, login as the <ra_user> and start the

RepAgent/MsSQL instance.

5. Log into RepAgent and configure the

connectivity properties described previously

(up above in step #5), though you can omit

pds_password.

6. Also configure the RepAgent

pds_integrated_security property to "true":

ra_config pds_integrated_security, true

By default, this property is set to "false".

7. Continue configuring/using RepAgent as

described in Replication Agent documentation.

fails when the package was originally created

with errors.

For direct connections to the repository, the customer uses NT Authentication.

They would like to use this same for the proxy server.

9/13/2006

This issue has been reviewed by engineering and has been closed with the status 'Not to be fixed' and the following explanation for this decision:

We currently do not intend to let the dabase connection user be changed with proxy.

There is only one db user that can be defined in the proxy configuration tool. We do not see why there is a need to deduce the db connection user from the NT user on the client. We do not see which advantadge the customer could have from thid.

Even if this is possible today with a direct connection, this is really not recommended....

Right management in repository is based on repository user not on db user.

The workaround is to search a specific CDM from the workspace browser

RA_OP_ID_ are created with precision 9 instead

of 19.

After applying this EBF, you must migrate each

existing Replication Agent instance that has

marked tables or procedures:

1 - Start the Replication Agent instance

2 - Login to the Replication Agent instance

and issue the following command:

pdb_xlog

After the instance migration is complete, you may

resume replication:

resume

handling maps system object ids prior to system

object id map completion.

the date in the 'Configure a Server Product'

option as 'Sep 18 106 17:54' rather than 'Sep 18

2006 17:54'.

"A commit is seen, but a multi-record

operation reference is still open"

multiple replication definitions may cause a deadlock cycle that spans

the ASE server and the RS server (i.e. an application deadlock). If the

ISQL script is interrupted with a Ctrl-C.

when user attempts to unmark an unmarked object

(table, procedure, sequence).

with error: LTI unable to process

unsupported operation: -1(UNDEFINED)

from update to large table.

certain type of commit log record.

rename option for an alter table DDL command.

terminates abnormally.

retrying connection after Replication Server

outage.

pdb_setreptable.

Description:

Allows one or more tables to be 'disabled' from

replication. Allows previously disabled tables to

be re-enabled, and allows the current status of

existing tables to be listed.

This command can be used to disable replication

of tables whose content need not be replicated.

Disabling replication in the Replication Agent

overrides the marking status established in the

primary Adaptive Server Enterprise database by

ASE commands sp_setreptable and sp_reptostandby.

Syntax:

To return replication enabled status:

pdb_setreptable {tablename|enable|disable}

To enable, or disable all marked tables:

pdb_setreptable all, {enable|disable}

To enable, or disable a specified table:

pdb_setreptable tablename, {enable|disable}

Parameters:

tablename

The name of a user table in the primary database.

The tablename parameter can be owner-qualified to

include the primary table owner name, with each

element separated by a period. For example:

owner.table

This parameter can be delimited with quote

characters to specify the character case.

If mixed character case (both uppercase and

lowercase) is required, the name must be

delimited. For example:

"Owner".table

"Owner"."Table"

all

A keyword that refers to all tables in the

primary database. By using the all keyword you

can apply an enable or disable operation to

all marked tables.

enable

A keyword that refers to enabling replication for

marked tables.

disable

A keyword that refers to disabling replication

for marked tables.

Usage notes:

If pdb_setreptable is invoked with no parameter,

it returns a list of all marked tables in the

primary database for which replication is

enabled. Note this is the same result returned

when the enable keyword is supplied.

If pdb_setreptable is invoked with a table name,

it returns complete marking information about

the specified primary table.

If pdb_setrepable is invoked with the enable

keyword, it returns a list of all marked tables

in the primary database for which replication is

enabled.

If pdb_setrepable is invoked with the disable

keyword, it returns a list of all marked tables

in the primary database for which replication is

disabled.

If pdb_setreptable is invoked with a table name

and the disable keyword, it disables future

replication of activity which occurs against

the specified marked table.

If pdb_setreptable is invoked with a table name

and the enable keyword, replication is re-enabled

for the specified marked table that was

previously disabled.

When pdb_setreptable is invoked with the all

keyword and an action keyword (enable or disable)

the action specified is applied to all marked

tables in the primary database.

If pdb_setreptable is invoked with the all and

disable keywords, it disables replication for all

marked tables in the primary database.

If pdb_setreptable is invoked with the all and

enable keywords, it enables replication for all

marked tables in the primary database that were

previously disabled.

Note that the output from command pdb_setreptable

contains several columns of information,

including the table's owner, name, mark and

enabled status. Only the result of the 'Enabled'

column can be changed using the pdb_setreptable

command. Other column values are only

informational.

Note that the output from command pdb_setreptable

contains some column values which mimic output

from other Replication Agents and their usage is

not applicable to ASE. The results from these

columns are returned as "N/A".

Examples:

pdb_setreptable authors

This command returns replication marking

information for the table named authors in the

primary database.

pdb_setreptable disable

This command returns replication marking

information for all marked tables for which

replication has been disabled.

pdb_setreptable all, enable

This command enables replication for all marked

tables in the primary database.

pdb_setreptable authors, enable

This command enables replication for the marked

table authors in the primary database.

incorrectly then resumes replication, the RepAgent

hangs.

large LOB processing can causes Replication Agent

to hang.

not find Resource Bundle containing index: XXX"

where XXX is a message key.

pdb_get_procedure_parms is missing. This version

of the command returns the list of all parameters

for a specific owner and procedure.

interrupting the RS server user thread with Ctrl-C will sometimes cause

the RS server to terminate immediately.

database is not replicating transactions

at the time a switch active is

initiated, when the switch active

completes, changes to the new standby

(former active) database is incorrectly

written into the inbound queue of the

logical connection.

the USA's change, effective in March 2007, as to

when Daylight Saving Time starts and stops. The

following JREs, which have the Daylight Saving Time

fix, also have a bug related to the use of the

JISAutoDetect character set:

Solaris SPARC, Linux, Windows: 1.4.2_12

HP PA-RISC: 1.4.2_11

As a workaround, RepAgent currently does not

support this character set on *any* platform.

The JISAutoDetect character set "Detects and

converts from Shift-JIS, EUC-JP, ISO 2022 JP

(conversion to Unicode only)".

ra.bat cannot be entered into Windows Service.

But Block# is returned to the correct position, when new transaction begins.

6146 errors will be caused if quiece_check is executed by the way in this state. But this error is mistake.

"Log reposition point locator"

has an invalid hex value indicator.

interval, due to the incorrect calls to the

non-thread safe routine 'localtime()'. Calls to

this routine are replaced with calls to the

thread safe routine 'localtime_r()'.

record that is generated when a table is altered

by dropping a column with data rows.

Repository server : MS SQL Server 2000

Target DBMS: DB2 8 common server

Proxy server: PDM with views fails to extract from migrated repository

Customer has upgraded from PD 11.x to PD 12.1.0 EBF2. Via a direct connection to the repository, they can consolidate and extract all models. However, via a proxy server connection, they get the following error message when extracting a PDM:

PowerDesigner � Error

A command sent to the repository proxy failed.

22005 (SQLSTATE [Microsoft] [ODBC SQL Server Driver] Invalid character value for cast specification)

When reviewing the proxy server log, the following warning is documented after this ODBC error:

ExtractModel: Failed when extracting OBJT = xxxxx

Upon querying the PMOBJT table, this object id belongs to a view

For repositories created in v12 or 12.1, this error does not occur.

ntext column results in the following error:

Failed to Open LOB data stream.

contains a LOB column is marked for replication

and the LOB column is enabled for replication, a

user other than the pds_username user gets an

error when inserting into the primary table.

IMPORTANT: This fix works only for tables that are

marked AFTER this EBF has been applied to

RepAgent. The WORKAROUND for existing marked

tables is the following:

1 In a text editor, open the mark.sql script for

the desired table. The script will be in:

$SYBASE/rax-12_5/<inst>/scripts/<tbl>/installed

where <inst> is the RepAgent instance name and

<tbl> is the table name

2 Find and copy the text for creating the shadow

row procedure to another file.

3 In the new file, edit the shadow row procedure

by owner-qualifying the name of the blob-shadow

row procedure as this snippet indicates:

BEFORE:

=======

SELECT @opid = @@IDENTITY

IF @dummy > 0

BEGIN

EXEC qa3bsrp_d @tranid,@opid,@intype

END

END -- not = 'a'

AFTER:

======

SELECT @opid = @@IDENTITY

IF @dummy > 0

BEGIN

EXEC <rausr>.qa3bsrp_d @tranid,@opid,@intype

END

END -- not = 'a'

where <rausr> is the name you configured in the

pds_username parameter, or possibly 'dbo' if

the <rausr> was granted full SA permissions.

4 Save this file.

5 In SQL Server, lock the primary table to prevent

any modifications (so replication won't lose any

operations or transactions against this table).

6 In SQL Server, drop the existing shadow row

procedure.

7 In SQL Server, re-create the shadow row

procedure using the saved SQL script you edited.

8 In SQL Server, unlock the primary table.

primary key includes more than one column is

incorrectly identified.

the drop column option for an alter table DDL

command.

primary key column contains single quotes,

RepAgent may go to ADMIN state with an error

similar to: Unclosed quotation mark before the

character string ''.

NOTE that to replicate character data that embeds

single quotes, RepAgent must be configured to use

structured tokens:

ra_config structured_tokens, true

is not correct. The same behavior can be observed with local merging.

processing alter table DDL command with rename

option.

NullPointerException instead of returning the

database's error message.

example: when maintenance user > 24 characters and dbname = 21 characters ...... the number of combined characters of the database and the maintenance user id is > 42

The display of this info as it is an Internal information, it should not been displayed specially when in Tools --> display preferences, Code is selected.

replication of inserts/updates/deletes from

Multidimensional Clustered Tables.

non-English(en_US) locale. It fails

with the following message

''Encountered an invalid build date.

Reason: Unparseable date: ''Fri Nov

17 19:08:41 MST 2006''.

with a NullPointerException when executing

a pdb_setrepproc command to mark a procedure

that contains no arguments.

processed the first time a Replication Agent

reaches replicating state.

performance.

an application deadlock with the ASE server when the Replication Server

is configured to use Parallel DSI and is applying transactions to a

table containing TEXT/IMAGE columns.

When this situation occurs, the Replication Server 'admin who' command

will display the Replication Server thread 'DSI EXEC' in either the

thread state 'Awaiting Commit Order' or 'Awaiting Batch Order'. The

thread state of the thread will not change over time.

To correct this issue, suspend and resume the Replication Server DSI

connection. If the Replication Server DSI thread fails to suspend,

re-start the Replication Server.

and never applied at databases replicated through subscriptions. They are not applied

at warm standby databases if the transaction is not a large transaction.

For large transaction replicating to warm standby, if the DSI is fast, the commands

in the mini rollback section may be applied at standby and then rolled back. These

commands may cause error at standby and cause DSI to shutdown. Users should resume

connection to clear out the error.

creates a repdef which adds columsn to the internal repdef. If the new repdef lists the

columns not in the order listed by the table definition, the column order DSI sends to

the warm standby or MSA will not match the prepared dynamic SQL statement. This could

replicate the data out of column order or suspend the DSI. The problem can be avoided by

suspending the DSI before the repdef is created.

than 2GB may fail depending on the size

of the partition. In some cases, error

6125 is generated after the failure. In

other cases, no error is generated after

the failure.

MICROSOFT SQL SERVER ONLY:

Repository code for mark/unmark.

Module(pdb_setreptable and related pdb_xxx

properties ,ra_helparticle,ra_helpfield...)