sp_downgrade_esd fails for ASE 15.7 SP134

[Former Member]

Anyone came across this issue and have a workaround?

sp_downgrade_esd from 15.7 SP 134 to 15.7 ESD#2 fails with error:

Attempt to remove an anchor for object 'sysdams' in database [] failed.

The error is not consistent - some DBs are downgraded just fine, some consistently fail. 

There is a seeming workaround in executing sp_downgrade_sysdams code manually (dbcc dam .. drop|create) but the command too has the same problem:  for some reason it fails to delete rows from system tables (syspartitions, sysindexes and syobjects) in correct manner so that the table sysdams is dropped.  As the result - the error message is generated, the table IS dropped and the catalog becomes corrupt (Inconsistent row anchor: object id 58 - for the three said tables).  If the sysdams is recreated manually, the corruption passes.

The problem occurs when the DBs are upgraded from the earlier release, used for some time, and than rolled back.  The same procedure worked fine for the databases that were freshly upgraded and rolled back.  No aync log service involved in this issue.  The downgrade process follows the downgrade procedure specified in documentation for the relevant platform (Installation Guide for Solaris 10).

Any ideas appreciated.

Thanks,

Andrew

[Former Member · Accepted solution]

Problem solved:  if system tables are bound to a named cache dbcc dam will fail with error.  unbind the system tables from cache before running sp_downgrade....

[Former Member]

Actually, the problem in my case is that the sysanchors is not cleaned up for some databases.

If you run the same dbcc dam drop for the database where the command works fine and check the content of sysanchors with dbcc anchors you see that the entries for sysdams are removed.  For those DBs where it fails - the entries remain.

[Former Member]

Can't help - hopefully Bret can help here.

But I had an issue with downgrade from SP132 - see here

http://scn.sap.com/thread/3707909

not the same error but could be similar.

[Former Member]

Thanks Mike.  You've encountered a different problem though.  I have copied the DB and dropped all user objects - still the same issue.  Hopefully TS will have a say on it.  Did not find anything relevant on KBA or SCN.  It looks, though, that something in the downgrade process is not functioning:  I've debugged the sp_downgrade_sysdams proc and found that when the proc tries to drop sysdams there still exist entries in syspartitions on it.  Either the dbcc command is meant to delete those or there is some other thing misfunctioning.  I'll have to wait for TS.