Re: ORA-01555 error

I don't have a reproducible test case currently, as this was a manual purge operation run by another dba.
I was called in when the existing undo space was exhausted. It was a real system, in a non-typical usage scenario. The "badness" was whereby space allocated to expired undo segments was not being "stolen" by online undo segments. Such scenarios have been documented by others elsewhere long ago.

Again, this was a single user system.
Ora-1555s were not being thrown.
The error condition was simply that undo segments were being brought online and space was not being released. The available space within the undo tablespace was exhausted. Oracle simply wanted more.

Reworking the statement into PL/SQL with commit logic driven off of a mod ( loop ) provided a work-around.
It was not worth filing a service request over, as it is not a scenario that is commonly encountered in our environment.

Bugs happen. corner cases happen. don't be surprised.

this was waaaay too proprietary to show a test case.

the reason that I posted was to backup the OP to state that shit happens.
It was not to post a reproducible test case of a failure mode that I don't care to pursue being diagnosed. If I did care enough, I'd pursue a service request.

-bdbafh Received on Wed Dec 07 2005 - 00:30:52 CST