[sheepdog-users] Impossible to run sheep on existing journal

[Liu Yuan]

On 12/18/2012 08:56 PM, Valerio Pachera wrote:
> sheep -j dir=/mnt/sdb1/sdj,size=512 -w object:size=20000 /mnt/sheepdog/
> 
> If I stop the daemon on this node and try to run it again later, it
> doesn't start.
> If I remove the content of the journal dir, it starts.
> 

How did you stop the daemon? collie kill node?

> From sheep.log:
> Dec 18 13:51:07 [main] jrnl_recover(230) opening the directory
> /mnt/sheepdog/journal/
> Dec 18 13:51:07 [main] jrnl_recover(235) starting journal recovery
> Dec 18 13:51:07 [main] jrnl_recover(291) journal recovery complete
> Dec 18 13:51:07 [main] send_join_request(1014) IPv4 ip:192.168.2.41 port:7000
> Dec 18 13:51:08 [main] replay_journal_entry(147) open No such file or directory

how to reproduce it reliably? would be nice if you can enable '-d' for
sheep daemon.

For the last resort, you can 1) remove all the log file and 2) restart
the sheep, then 3) collie vdi check the images.

Thanks,
Yuan