[sheepdog-users] 0.8.0 crash

[Liu Yuan]

On Fri, Mar 14, 2014 at 01:32:32PM +0100, richter at ecos.de wrote:
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Hi,
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> here another crash of 0.8.0. After a few restarts it starts recovery.
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Regards
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Gerald
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> Mar 14 13:22:31
> 
>  WARN [gway 83987] wait_forward_request(413) poll timeout 1, disks of some
> nodes or network is busy. Going to poll-wait again
> 
> 
> 
> 
> 
> 
> 
> Mar 14 13:22:57
> 
>  WARN [gway 84020] wait_forward_request(413) poll timeout 1, disks of some
> nodes or network is busy. Going to poll-wait again
> 
> 
> 
> 
> 
> 
> 
> Mar 14 13:23:04
> 
>  WARN [gway 83985] wait_forward_request(413) poll timeout 1, disks of some
> nodes or network is busy. Going to poll-wait again
> 
> 
> 
> 
> 
> 
> 
> Mar 14 13:23:54
>  ERROR [net 58458] do_read(236) failed to read from socket: -1, Connection
> reset by peer
> 
> 
> 
> 
> 
> 
> 
> Mar 14 13:23:54
>  ERROR [net 58458] rx_work(742) failed to read data
> 
> 
> 
> 
> 
> 
> 
> Mar 14 13:23:54
>  EMERG [gway 84036] add_to_lru_cache(693) PANIC: the object already exist
> 

Probably there is some corner case for object cache. You can rm /cache/* -rf
before restart to work around it.

Thanks
Yuan