Known Issues for Replication Server

Workaround: Use sysadmin system_version to set the ID server system version to 1260 or later, and then restart the preupgrade Replication Server. You can also use sysadmin system_version to check the system version at the preupgrade Replication Server.

• Greater than 32K, you see "Incorrect syntax near '?'" and the DSI thread shuts down.
• Less than 32K and the column has compressed Adaptive Server data, this data become corrupted in the replicate database.

• Mark the in-row LOB columns that are to be replicated as always_replicate instead of replicate_if_changed.
• Do not use the in-row LOB feature in Adaptive Server (that is to set in-row length to 0) when the column is to be replicated.
• Set dsi_compile_enable off to disable HVAR on the replicate Replication Server.

If these workarounds do not resolve the issue, contact Sybase Technical Support to obtain an EBF.

Although the maximum allowable size is 2,251,799,813,685,247, the actual maximum value is 2,147,483,647 (2GB -1).

Workaround: Alter the replication definition to match the table schema.

Workaround: Do not use both features simultaneously.

When you upgrade a route from pre-15.7 versions to version 15.7 in a mixed-replication environment, the alternate connection at the primary Replication Server is not populated into the replicate Replication Server RSSD.

1. Drop the alternate connection.
2. Re-create the alternate connection.

Workaround: Use sp_addlogin to add the maintenance user to the standby database.

Workaround: Define the blob column as nullable:

• In the database, and,
• If you have a table-level replication environment, the replication definition created for the blob table replication.

Workaround: Manually start the sample Replication Server after installation.

These steps result in some updates replicating incorrectly:

Workaround: Drop and re-create the partitions that were created before the block size change.

Workaround: Issue this trace option in rs_init:

rs_init -r resource_file_name -T T_SEND_CLEARTEXT_PASSWORD

In a warm standby environment, if you set text and image columns of a table to do_not_replicate on the active database, and you create a table replication definition for this table without send standby replication definition columns, the replication definition does not include the text and image columns. As a result, the standby database gets incorrect value for the text and image columns.

1. Create the table replication definition with send standby replication definition columns clause.
2. Drop the table replication definition if it is not necessary.

• All Replication Servers involved in the previous topology are quiesced. Verify with admin quiesce_check.
• RepAgent starts reading from the end of the primary database log.

If you create multiple replication definitions for a specified primary table and you subsequently alter a column without an LOB datatype to be an LOB column in one of the replication definitions, the LOB-related function strings are not created automatically for one of the replication definitions at the replicate Replication Server. The replicate Replication Server log shows:

Cannot find a matching function string for function 'alltypes.rs_writetext' and function string class 
'rs_sqlserver_function_class'.

1. Clear the failed Data Server Interface (DSI) connection queue by executing resume/skip tran until the DSI connection resumes.
2. Drop the subscription to the problem replication definition at the replicate Replication Server, then drop the replication definition at the primary Replication Server, then re-create both.

If a Stable Queue Transaction (SQT), Stable Queue Manager (SQM) page caches, or DSI Bulk/HVAR buffers are using heavy memory loads, Replication Server may fail when it exceeds the low default ulimit settings in the operating system.

See your operating system documentation for similar implementations.

Message from server: Message: 2601, State 6, Severity 14 -- 'Attempt to insert duplicate key row in object 'rs_translation' with unique index 'rs_key_translation'

The rs_translation system table has unique index on (classid, source_dtid). The two translation (ASE binary to rs_oracle_binary and ASE timestamp to rs_oracle_binary) generates the duplicate key insert error.

Workaround: For Replication Server versions 15.5 and later, you have to manually remove the translations when you encounter this error. Open the SQL scripts and search for timestamp to find the related SQL statements.

delete from rs_profdetail where profid = 0x000000000e000000 and pdetailid = 0x000000000e00003f
delete from rs_systext where parentid = 0x000000000e00003f and sequence = 1
delete from rs_profdetail where profid = 0x000000000e000000 and pdetailid = 0x000000000e000040
delete from rs_systext where parentid = 0x000000000e000040 and sequence = 1

delete from rs_profdetail where profid = 0x000000000e010000 and pdetailid = 0x000000000e010042
delete from rs_systext where parentid = 0x000000000e010042 and sequence = 1
delete from rs_profdetail where profid = 0x000000000e010000 and pdetailid = 0x000000000e010043
delete from rs_systext where parentid = 0x000000000e010043 and sequence = 1

delete from rs_profdetail where profid = 0x000000000e020000 and pdetailid = 0x000000000e02003f
delete from rs_systext where parentid = 0x000000000e02003f and sequence = 1
delete from rs_profdetail where profid = 0x000000000e020000 and pdetailid = 0x000000000e020040
delete from rs_systext where parentid = 0x000000000e020040 and sequence = 1

delete from rs_profdetail where profid = 0x000000000e00000c and pdetailid = 0x000000000e000c27
delete from rs_systext where parentid = 0x000000000e000c27 and sequence = 1
delete from rs_profdetail where profid = 0x000000000e00000c and pdetailid = 0x000000000e000c28
delete from rs_systext where parentid = 0x000000000e000c28 and sequence = 1

delete from rs_profdetail where profid = 0x000000000e000012 and pdetailid = 0x000000000e001231
delete from rs_systext where parentid = 0x000000000e001231 and sequence = 1
delete from rs_profdetail where profid = 0x000000000e000012 and pdetailid = 0x000000000e001232
delete from rs_systext where parentid = 0x000000000e001232 and sequence = 1

Workaround: Either:

If you use isql and SYBASE.csh to connect to Replication Server through SSL, the connection fails. isql does not initialize the network filter.

source SYBASE.csh
setenv LIBPATH $SYBASE/$SYBASE_OCS/lib3p:$LIBPATH

source SYBASE.sh
export LIBPATH=$SYBASE/SYBASE_OCS/lib3p:$LIBPATH

If the libsybdldap.so and libsybdldap64.so are not specified in the Directory section in libtcl.cfg, Replication Server cannot connect to the LDAP.

For example:

[DIRECTORY]
ldap32=libsybdldap.so ldap://sylvester:3389
/dc=Sybase,dc=com??one??bindname=cn=Manager,dc=Sybase,dc=com??secret  
ldap64=libsybdldap64.so ldap://sylvester:3389
/dc=Sybase,dc=com??one??bindname=cn=Manager,dc=Sybase,dc=com??secret

The entry name has changed from ldap to ldap32. To start the LDAP session with dscp utility:

open ldap32

When configuring Replication Server 15.5 or later using the rs_init with an existing RSSD, an error occurs in rs_init.

If you increase the number of client connections and did not increase the number of Open Server™ threads that the Replication Server can use, Replication Server may shut down.

1. Log in to RSSD.
2. Increase the value of num_threads.
3. Restart Replication Server.

If the sort order is not set correctly in the [collate] section of $SYBASE/config/objectid.dat, a stack trace error occurs during Replication Server start-up.

1. Make sure that the character set and set order have been set correctly in Rep_Server_name.cfg file. For example:
   • RS_charset=cp850
   • RS_sortorder=scannocp
2. Go to $SYBASE/charsets/character set; for example, $SYBASE/charsets/cp850.
3. Locate and open corresponding sort order file; for example, scannocp.srt.
4. Check the line that has an “id” and find the sort order ID, for example,

   id = 0x30; Unique ID # (48) for the sort order

   The sort order ID is 48.

5. In $SYBASE/config/objectid.dat, check that there is a line for that sort order under the [collate] section:

   1.3.6.1.4.1.897.4.9.3.48 = scannocp

   The last number is the sort order ID (48), all other numbers are identical for all lines. If you have a custom sort order, add this line for it.

6. Restart Replication Server.

This may impact your replication performance.

Workaround:

• Use single quote (‘) or double quote(“) around the long repdef_name.
• Use a truncated repdef_name. Truncate the repdef_name to the first 29 characters when issuing the rs_helprep request. The rs_helprep then appends the % wildcard at the end when it queries the RSSD tables.
• Upgrade RSSD data server to Adaptive Server 15.x.

Workaround: Create the subscription without a quoted constant or create the subscription without materialization.

Workaround: Suspend and resume the DSI connection. If the DSI thread does not suspend, restart the Replication Server.