RE: What can cause a rapid frequency of log switches?
- From: "Mark W. Farnham" <mwf@xxxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 4 Jun 2004 18:56:20 -0400
Are the archived logs about 400MB, or are they very short or zero length?
If about 400MB, then something is causing a lot of redo generation. If the
number of archives is getting to be bothersome, make them bigger.
If zero or about zero, then it *might* be failing to write to the next group
so it switches. If it can't find something valid to switch to, this should
be pretty obvious, since commits will hang until you can write them to redo.
I can't remember whether it marks a group invalid if it fails to write to it
or whether it could possibly run in a circle until you give it something
usable to write to. Failures would almost certainly be found in the alert
log.
mwf
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx]On Behalf Of Jacques Kilchoer
Sent: Friday, June 04, 2004 6:24 PM
To: oracle-l@xxxxxxxxxxxxx
Subject: RE: What can cause a rapid frequency of log switches?
Not my database though and I'm not at the site. I'm telling my friend "it
has to be database activity, there is no other reason."
----------------------------------------------------------------------------
--
Jared.Still@xxxxxxxxxxx
oracle-l-bounce@xxxxxxxxxxxxx wrote on 06/04/2004 01:38:23 PM:
> Oracle 9.2.0.4 / AIX 5.1
>
> I'm looking at an issue with redo log switches.
> As far as I know, a log switch will only happen
> A) if the log is full
> B) if someone enters an "alter system switch logfile"
> C) if archive_log_target is set
>
> I've been asked why there would be a log switch in less than a minute
> (the redo logs are all 400 MB in size). The only answer I could think of
> is that there is a lot of activity in the database. Is there some
> possibility I'm overlooking?
>
Well, you could speculate, or you could use log miner and see
what's going on.
Jared
Other related posts:
