[sheepdog-users] object recover each time cluster startup

[Liu Yuan]

On Fri, Aug 30, 2013 at 12:14:25PM +0900, MORITA Kazutaka wrote:
> At Thu, 29 Aug 2013 12:08:16 +0200,
> Valerio Pachera wrote:
> > 
> > I shutdown the cluster by 'dog cluster shutdown'.
> > No guests are running obviously.
> > 
> > When I restart sheep I get
> > 
> > Aug 29 12:04:06   INFO [main] md_add_disk(141) /mnt/sheep/dsk01/obj, nr 1
> > Aug 29 12:04:06   INFO [main] md_add_disk(141) /mnt/sheep/dsk02, nr 2
> > Aug 29 12:04:06   INFO [main] send_join_request(770) IPv4
> > ip:192.168.2.47 port:7000
> > Aug 29 12:04:06  ERROR [main] for_each_object_in_stale(383)
> > /mnt/sheep/dsk01/obj/.stale
> > Aug 29 12:04:06  ERROR [main] for_each_object_in_stale(383)
> > /mnt/sheep/dsk02/.stale
> > Aug 29 12:04:06   INFO [main] main(821) sheepdog daemon (version
> > 0.7.0_50_g8f659a2) started
> > Aug 29 12:04:07  ALERT [main] get_vdi_copy_number(79) No VDI copy
> > entry for 98390b found
> > Aug 29 12:04:07  ALERT [main] get_vdi_copy_number(79) No VDI copy
> > entry for 98390b found
> > Aug 29 12:04:07  ALERT [main] get_vdi_copy_number(79) No VDI copy
> > entry for 98390b found
> > <cut>
> > 
> > in sheeo.log of all nodes.
> > 
> > Any idea?
> 
> The error message "No VDI copy entry for xxxx found" will not be shown
> with the current master, I think.
> 
> Note that Sheepdog starts object recovery even after clean shutdown.
> 

I think we shouldn't start recovery for a clean shutdown. I've sent the patch
to fix this.

Thanks
Yuan