Re: select for update wait issues
From: Mladen Gogala <mgogala_at_yahoo.com>
Date: Mon, 13 Apr 2015 17:30:10 -0400
Message-ID: <552C3562.6000000_at_yahoo.com>
On 04/13/2015 01:14 PM, Sandra Becker wrote:
> Several waits:
>
> EVENT ROW_CNT
> ----------------------------------- ----------------
> db file sequential read 7,714
> enq: PR - contention 551
> latch free 11,020
> log file sync 96,425
> read by other session 7,163
>
Date: Mon, 13 Apr 2015 17:30:10 -0400
Message-ID: <552C3562.6000000_at_yahoo.com>
On 04/13/2015 01:14 PM, Sandra Becker wrote:
> Several waits:
>
> EVENT ROW_CNT
> ----------------------------------- ----------------
> db file sequential read 7,714
> enq: PR - contention 551
> latch free 11,020
> log file sync 96,425
> read by other session 7,163
>
Sandra, how did you extract those wait events? What does the column ROW_CNT mean? The only meaningful value is the time spent waiting. You should extract the wait events from V$SESSION_EVENT table, for the session that is doing "SELECT FOR UPDATE". The meaning of the the wait events above is:
- Index read of a table
- Start process
- Latch free (cannot be described in more details without knowing which latch it is)
- Commit processing
- Self explanatory, probably related to latch waits.
If I was allowed to wild guess the problem, I would say that there is an index range scan where there hasn't been one before and that the table is frequently used, something like addresses, phones or ZIP codes.
-- Mladen Gogala Oracle DBA http://mgogala.freehostia.com -- http://www.freelists.org/webpage/oracle-lReceived on Mon Apr 13 2015 - 23:30:10 CEST