[Pgcluster-general] FYI: PGCluster-1.9.0rc1 released
Rick Vernam
rickv at hobi.com
Wed Feb 6 16:22:21 UTC 2008
On Tuesday 05 February 2008 10:10:01 pm At.Mitani wrote:
> Hi Rick,
>
> > I've written against 1.7-whatever..
> > If you plan to include in 1.9-whatever, I'll supply a patch against
> > 1.9.0rc1
> >
> > do you want for 1.5 & 1.3 as well?
>
> Great!
> I'm looking forward to get your patches :)
>
> > > By the way,
> > > When rsync becomes time-out, the recovery function will always fail as
> > > you know. This is a fatal flaw for the HA system, I think.
> >
> > Indeed, I agree. But don't you think that ignoring a failing sub-system
> > is even worse?
>
> Yes, I agree.
> There is no doubt that ignoring a failing sub-system is needed.
>
> On the other hands,
> I want to make recovery succeed even in the environment which fails rsync,
> if possible.
For now, I'll just provide for a user defined bwlimit & timeout.
Perhaps in the future, we could have the recovery feature allow rsync to
timeout after no activity, but then retry where it left off with some
bwlimit. So, for example, no bwlimit on first attempt, second attempt it
could use a user-defined bwlimit, and if that fails make a third attempt with
some lower bwlimit...
>
> Therefore, I'm looking for a possibility, not only the rysnc options but
> also other tools...
>
> Of course, I'm great hopes in your patches.
>
> Regards,
More information about the Pgcluster-general
mailing list