Re: ORA-03113 - end-of-file on communication channel

rwessman_at_us.oracle.com wrote in message ...
>
>That said, if the fault is not caused by a server bug (check the RDBMS log
>directory for trace files), a 3113 can be hard to track down.

Agree. I had 3113 on a SQL statement that created a fairly large (400+MB) spool file. Bombed out after about 100MB or so with a 3113. Tried everything and anything to get the run through without any luck. Finally in desperation I changed script from using a local connection to a SQL*Net connection - and there it worked!

Oracle's only comment was that it must be some kind of bug and left it at that as they could not reproduce it all.

The other times I got the problem it was due to a bug in the Unix protocol stack - at least I think it is as it disappeared after we did an operating system upgrade. It can also be the local network driver. After upgrading the SMC drivers on a PC that was also getting this error it also seemed to disappear.

Weird. A very intermittent error. Sometimes we got it a quite a few times a day and other times a week passed with maybe one or two 3113 errors. But this problem disappeared after we did several upgrades of the operating system and took Oracle from 7.2 to 7.3 and applied various 7.3 patches.

My 2'c worth of advice. Make sure that you have applied all the listener and SQL*Net patches for your platform when you get an 3113. Also have a look at DCD (dead connection detection) if you have it enabled as it was also buggy for quite a while.

regards,
Billy Received on Tue Oct 27 1998 - 09:01:30 CST