Hi, I have following ASE 15.7 SP134 on Linux.
a stored procedure using tempdb is failing with tipical 1105 error (system segment is full).
tempdb is 16GB and stored procedure use around 14.5Gb leaving 1.5GB free when we get the error.
the transaction disappear, the # table disappear but the space remains in use as showed by sp_helpdb tempdb output.
dbcc checkalloc(tempdb) shows a wide multiple corruption with error Msg 2546.
After reboot we expect tempdb is free of corruptions but dbcc checkalloc(tempdb) output shows are still there: corruptions are persistent over reboot.
We reboot several times without success.
We deleted tempdb devices reducing tempdb to the default 'master' size, removed files and recreated devices and increased tempdb as before, corruptions went away but running again the stored procedure the corruptions appeared again and again reboot did not fix it.
No errors in error log (no stack trace)
No errors in Linux log.
No errors in storage.
tempdb is on filesystem as the other user databases on the same filesystem and path, the other databases do not have any problem.
The other devices have 'direct IO on', tempdb has direct io off and dsync off as well, turning direct io on (and rebooting) corruption does not go away.
There are other ASE with identical version and configuration and databases on the same machine but they does not show any problem.
Cannot find CR/bug related to this issue in SAP support pages.
I do not know what to check anymore!
Question: can you explain how it is possible that after reboot tempdb still have corruption ? If it is 'created' should not have any corruption !
P.S. model if free of corruption dbcc checalloc and checkdb does not show any error.
Any Idea ? Any check to propose ?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.