Re: DBWR timed-out, what causes this?

In article <917469260.2590.0.nnrp-02.9e984b29_at_news.demon.co.uk>, "Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> writes:
>It would be worth checking the hardware in case one disc
>is throwing large numbers of write errors, causing writers
>to keep re-trying.

I checked, there are no errors on any of the disks.

This is our current thought -- perhaps the time-out value on the physical disk is too low? It will be raised today from ~100 to 300.

Is there any way to change the time-out value for the database writer process?

>Also, are there any trace files dumped for the slaves that
>might give a clue.

Unfortunately it was the master DBWR that died. No slave traces. PMON also dropped a trace stating that DBWR had died.

>The other thing worth doing is finding our the session ids for
>the slave writers when the database starts, and monitor
>v$session_event for those SIDs, taking change values
>every 10 minutes and looking for extreme values.

I will give this a try. Thanks!

Rose

>>skubiszewski_at_Eisner.DECUS.Org wrote in message ...
>>The database died Monday night and the only clue I have
>>is a trace file from the DBWR with ORA-7404: sfareq:
>>Time-out occurred waiting for request to complete.
>>Cause: The master database writer timed out waiting
>>for a write or close to complete. One of the database
>>writers may have stopped running.
Received on Thu Jan 28 1999 - 09:01:55 CST