lio bod liobod.slony at gmail.com
Thu Feb 28 00:34:33 PST 2008
ok.I can see clearly now.
tfyt.




2008/2/27, Christopher Browne <cbbrowne at ca.afilias.info>:
>
> "lio bod" <liobod.slony at gmail.com> writes:
> > 2008/2/27, Christopher Browne <cbbrowne at ca.afilias.info>:
> > Agree. It's always obvious once it's written.
> >
> >
> >              - It is essential that a slon be kept running against the
> master
> >           node, as if it is not, for a period of time, then when it
> >           is restarted, that entire period of time will be treated as
> >           One Big Long SYNC, and having such a huge SYNC is likely to
> >           turn out badly :-(.
> >
> >
> >
> >
> > I don't catch the point
> >
> > The big sync may also turn badly if the slon 2 start and restart after a
> long time occures on the salve node.
> >
> > ?
> >
> > Thx anyway
>
> No, the point is that if the slon for node #1 was running, throughout,
> there won't be one single "very large SYNC."
>
> If the slon for the master node *is* running, then there should be a
> SYNC generated at least once every 2 minutes.
>
> Thus, if the slave DB is down for the weekend (e.g. -
> Saturday/Sunday), then when you return on Monday, you'll find that the
> slave node is behind by 2 days, and that the changes are divided up
> across, um...  At least 1440 SYNCs (30 per hour, times 2x24).  If
> there were times of heavy updates, there will more SYNCs than that.
>
> In the catch-up process, the subscriber slon, on Monday, will be able
> to gradually go through those changes, doing a COMMIT every so often,
> so that the changes from the weekend will gradually flow in.
>
> In contrast, if the slon for the "master" node was down for the
> weekend, as well, then there will be just 1 SYNC, covering the time
> from when the slon fell over until when it restarted.  There is no
> ability to do the updates in more "bite-sized" pieces.
> --
> let name=3D"cbbrowne" and tld=3D"cbbrowne.com" in String.concat "@"
> [name;tld];;
> http://linuxfinances.info/info/linuxdistributions.html
> Q: What does the function NULL do?
> A: The function NULL tests whether or not its argument is NIL or not.  If
>   its argument is NIL the value of NULL is NIL.
> -- Ken Tracton, Programmer's Guide to Lisp, page 73.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20080228/=
2bae4368/attachment.htm


More information about the Slony1-general mailing list