Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Rollback segments, ORA-01581
Hi Group,
I experienced problems with the rollback segements now and then in the
past but shrugged it off since nothing of significance happened to the
calling application. But for some days the error ORA-01581 starts
popping up for about every update/delete command I issue.
I know the purpose of rollback segments, I know they have to be big to
handle long transactions such as batch runs etc. Our rollback segments
are ***HUGE*** (about one third of the database size), they are set to
autoextend, unlimited extents and the database is configured to
"unlimited_rollback_segments = true". And there is PLENTY of diskspace.
It's Oracle workgroup server 7.3.3.4 on NT 4.0 SP 3, PII-200, Raid 5
Disk Array.
Today ORA-01581 came up when a user tried an update of 75 rows -
ridiculous. So I started a very big update and watched the storage
managers statistics. The point is: Nothing happened to the size of the
rollback segments or their datafiles sizes (Pl dont ask, Yes, I did
refresh the display expicitly about every 10 secs!). Just when the error
occurs the value "high water mark" will be greater than the value
"size", the affected rollback segment will change its status from
"online" to "full" and that's it. The calling application will throw an
exception (or the SQL Worksheet show the error), then the transaction
will be rolled back. If I repeat this update a few times other rollback
segments will be affected.
Thank You for hearing me out, now the question: Considering the described configuration, WHY won't Oracle claim diskspace as needed for the rollback segments and their datafiles??? Any help appreciated!
TIA Ralph Received on Wed Nov 17 1999 - 18:43:53 CST