RE: Question about log file sync in DataGuard
- From: Michael Schmitt <mschmitt@xxxxxxxxxxxx>
- To: David Fitzjarrell <oratune@xxxxxxxxx>, "oracle-l@xxxxxxxxxxxxx" <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 12 Oct 2012 19:45:28 +0000
Hi David,
Yes the standby redo logs are necessary on the standby server in 11.1.0.7 as
well. Since we are in LGWR SYNC mode there is no doubt some of our time is
waiting on the network and also the file I/O on the standby server. I was
suspecting this time was represented mainly by "LNS wait on SENDREQ". One of
the things I was unsure about, is if "LNS wait on SENDREQ" is just a subset of
"LGWR-LNS wait on channel"
Thanks,
Mike
From: David Fitzjarrell [mailto:oratune@xxxxxxxxx]
Sent: Friday, October 12, 2012 2:37 PM
To: Michael Schmitt; oracle-l@xxxxxxxxxxxxx
Subject: Re: Question about log file sync in DataGuard
I'm not sure about 11.1.0.7 but in 11.2 it's necessary to have standby redo
logs on the standby server; the waits could be network related as Oracle syncs
not only the local redo logs but also the remote logs on the standby.
David Fitzjarrell
From: Michael Schmitt <mschmitt@xxxxxxxxxxxx<mailto:mschmitt@xxxxxxxxxxxx>>
To: "oracle-l@xxxxxxxxxxxxx<mailto:oracle-l@xxxxxxxxxxxxx>"
<oracle-l@xxxxxxxxxxxxx<mailto:oracle-l@xxxxxxxxxxxxx>>
Sent: Friday, October 12, 2012 12:33 PM
Subject: Question about log file sync in DataGuard
Hello,
I am hoping the list can help me better understand the waits I am seeing for
log file sync in a Data Guard environment.
This is for an 11.1.0.7 database running on Linux and running in max
availability mode (LGWR SYNC)
We consistently see "log file sync" as the top wait event in AWR reports taking
30-40% of the DB time.
A two hour report today showed 1,938 seconds for log file sync.
The report also indicates the following background wait events.
LGWR-LNS wait on channel: 707(s)
LNS wait on SENDREQ: 680(s)
log file parallel write: 667(s)
I am trying to get a better understanding of the break down on the "log file
sync". My understanding is that 667(s) of the "log file sync" time is spent on
"log file parallel write" waiting for the LGWR on the primary database to write
to the redo logfile on the primary server. That would leave ~ 1271(s) time of
the reported "log file sync" left. Are the "LGWR-LNS wait on channel" and "LNS
want on SENDREQ" basically reporting on the same thing (sending data to standby
and waiting for confirmation) or are those two distinct functions? I was
thinking that they might be the same function and that approximately 564(s) of
the "log file sync" time was just multiple sessions reporting the "log file
sync" wait.
Any advice on better understanding this would be appreciated
Mike
--
//www.freelists.org/webpage/oracle-l
--
//www.freelists.org/webpage/oracle-l
Other related posts:
