[Pgcluster-general] new recovery rsync options, pgcluster

[Rick Vernam]

On Monday 21 January 2008 06:38:35 pm At.Mitani wrote:
> Hi Rick,
>
> It's an interesting configration.
>
> During recovery process, the rsync command is doing twice as you know.
> The first rsync is doing background as preparation.
> And the second rsync is doing foreground in order to lock each cluster DBs.
Over a long-distance, high-latency link with not-so-great NICs, rsync will 
hang indefinitely during the first stage if I don't set bwlimit to something 
low - between 10-15.  And if it does end up hanging, I like that it fails 
nicely due to timeout.

>
> Therefore, I'd like to recommend to use this setting at the second rsync.
> How do you feel?
I've seen rsync hang as much during first stage as I've seen during second 
stage.

>
> Regards,
> ---------------
> At.Mitani
>
>
> On Wed, 16 Jan 2008 22:48:11 -0600
>
> Rick Vernam <rickv at hobi.com> wrote:
> > I've added the Rsync_Timeout and Rsync_BWLimit configuration to pgcluster
> > 1.7-HEAD
> >
> > see
> > http://www.pgcluster.org/svn/pgcluster-1.7-HEAD/src/backend/libpq/cluster
> >.conf.sample for configuration details.
> > _______________________________________________
> > Pgcluster-general mailing list
> > Pgcluster-general at pgfoundry.org
> > http://pgfoundry.org/mailman/listinfo/pgcluster-general