[sheepdog-users] Testing snapshot-object-reclaim

[Hitoshi Mitake]

At Wed, 05 Mar 2014 20:40:36 +0900,
Hitoshi Mitake wrote:
> 
> At Wed, 5 Mar 2014 09:05:26 +0100,
> Valerio Pachera wrote:
> > 
> > I got a strange behaviour: after removing a big file on the guest, it
> > wasn't able to write a second one (no space left on the device).
> > Sheepdog daemon version 0.8.0_118_g1c75b2a
> > 
> > ---
> > dog vdi list
> >   Name        Id    Size    Used  Shared    Creation time   VDI id  Copies  Tag
> >   test         0   10 GB  2.3 GB  0.0 MB 2014-03-04 17:23   7c2b25
> > 2:1
> > 
> > dog vdi snapshot test
> > 
> > dog vdi list
> >   Name        Id    Size    Used  Shared    Creation time   VDI id  Copies  Tag
> > s test         1   10 GB  2.3 GB  0.0 MB 2014-03-04 17:23   7c2b25
> > 2:1
> >   test         0   10 GB  0.0 MB  2.3 GB 2014-03-05 08:50   7c2b26
> > 2:1
> > 
> > dog node info
> > Id      Size    Used    Avail   Use%
> >  0      233 GB  1.2 GB  231 GB    0%
> >  1      452 GB  1.2 GB  451 GB    0%
> >  2      500 GB  1.2 GB  499 GB    0%
> > Total   1.2 TB  3.6 GB  1.2 TB    0%
> > Total virtual image size        10 GB
> > 
> > <write big file on guest>
> > dog vdi snapshot test
> > <delete big file on guest>
> > dog vdi snapshot test
> > <write another big file on guest>
> > <error: no space left on the device>
> 
> It would be an error produced by the guest OS and its process, right?
> I want to see the error messages of commands and an output of dmesg if
> it is available.

Could you run sheeps with -l level=debug? I think this problem is
related to discard operation, so I want you to check whether the guest
OS issued discard requests or not.

BTW, which file system is your guest using?

Thanks,
Hitoshi