Re: intermittent long "log file sync" waits
From: Priit Piipuu <priit.piipuu_at_gmail.com>
Date: Fri, 7 Feb 2020 16:41:02 +0100
Message-ID: <CAJYY02g61BqWD2Qq7nU-VLNp9J64knoVxt0kKWpa73uLq7YFDA_at_mail.gmail.com>
Date: Fri, 7 Feb 2020 16:41:02 +0100
Message-ID: <CAJYY02g61BqWD2Qq7nU-VLNp9J64knoVxt0kKWpa73uLq7YFDA_at_mail.gmail.com>
Hi,
19c RAC and "log file sync" without associated dbwr activity rings a bell. I noticed similar thing while investigating DRM related problems. In our case GRD gets locked for a few seconds during remastering, and that totally wrecks certain DML heavy workloads. But it looked like a collateral damage and I didn't bother to investigate further. Of course, with remastering going wrong there would be "gcs drm freeze in server mode" wait events, and probably some other gcs related stuff from "Other" wait class. That's not apparent from the traces below.
On Tue, 28 Jan 2020 at 17:11, Chris Stephens <cstephens16_at_gmail.com> wrote:
> 3-node Oracle 19.3 RAC
> Centos 7
>
>
>
-- http://www.freelists.org/webpage/oracle-lReceived on Fri Feb 07 2020 - 16:41:02 CET