[sheepdog-users] update 0.8.0

Hitoshi Mitake mitake.hitoshi at lab.ntt.co.jp
Tue Aug 5 10:23:26 CEST 2014


At Tue, 05 Aug 2014 12:16:28 +0400,
Maxim Terletskiy wrote:
> 
> Thank you for reply!
> 
> One more question. I started to notice that amount of data I see via 
> 'dog vdi list' much less than ondisk used space(e.g. dog telling its 
> 3Tb, but in fact it's about 10Tb used). We're frequently using 
> snapshots, so this it maybe because of deleted snapshots space reclaim 
> problem in 0.8.0. Will space return back after update?

Unfortunately, the new object reclaim algorithm cannot be backported
to v0.8.x because it  completely changes the protocol of sheepdog.

The problem can be solved in v0.9 release. It would be released in the
near future.

Thanks,
Hitoshi

> 
> 05.08.2014 9:19, Hitoshi Mitake пишет:
> > At Mon, 04 Aug 2014 12:53:50 +0400,
> > Maxim Terletskiy wrote:
> >> Hi!
> >>
> >> I have cluster running 0.8.0. I'm going to add some more nodes and I
> >> want to update sheep at my nodes. So I have two questions:
> >>
> >> - What version of sheep choose to update to?
> > You should use v0.8.2:
> > https://github.com/sheepdog/sheepdog/releases/tag/v0.8.2
> >
> >> - Can I install new sheep version on new nodes before updating my old nodes?
> > Yes, old nodes and new nodes can coexist in single cluster if their
> > major version numbers are same (e.g. v0.8.0 and v0.8.2 can coexist).
> >
> > But please be careful that updating from v0.8.0 to v0.8.2 requires -u
> > option of sheep daemon. You need to add -u when you boot v0.8.2 sheep
> > once.
> >
> > Thanks,
> > Hitoshi
> 
> -- 
> С уважением,
> Терлецкий Максим
> ООО "ТБ"
> +7(499)917-0115
> http://www.emu.ru
> 
> -- 
> sheepdog-users mailing lists
> sheepdog-users at lists.wpkg.org
> http://lists.wpkg.org/mailman/listinfo/sheepdog-users



More information about the sheepdog-users mailing list