[Sheepdog] sheepdog on ubuntu 11.04

[MORITA Kazutaka]

At Fri, 29 Apr 2011 17:15:43 +0200,
S. Bonnegent wrote:
> 
> Hi,
> 
> I try to test Sheepdog on Ubuntu 11.04 64 bits with 2 hosts but I have
> some troubles to debug the situation.
> 
> I use 2 nodes (PC1 and PC2) with default sheepdog package
> (0.2.2-0ubuntu1). I added "user_xattr" (my partition is in ext4) in
> /etc/fstab and configured  /etc/corosync/corosync.conf (file is below).
> 
> PC1 is 172.29.22.74
> PC2 is 172.29.22.78
> 
> On PC1, I start sheepdog with "service sheepdog start" and I have:
> 
> # collie node list
>   Idx   Node id (FNV-1a) - Host:Port
> ------------------------------------------------
> * 0     52b76e70de45e6c8 - 172.29.22.74:7000
> 
> # corosync-cpgtool
> Group Name             PID         Node ID
> sheepdog
>                       1070      1242963372 (172.29.22.74)
> 
> 
> On PC2, I start sheepdog too and I obtain:
> # collie node list
> The node had failed to join sheepdog
> failed to get node list
> 
> # corosync-cpgtool
> Group Name             PID         Node ID
> sheepdog
>                       1070      1242963372 (172.29.22.74)
>                       1064      1310072236 (172.29.22.78)
> 
> And now, on PC1 I have:
> # collie node list
> The node had failed to join sheepdog
> failed to get node list
> 
> # corosync-cpgtool
> Group Name             PID         Node ID
> sheepdog
>                       1070      1242963372 (172.29.22.74)
>                       1064      1310072236 (172.29.22.78)
> 
> In PC1 logs, there are this message:
> 
> get_cluster_status(362) joining node has invalid ctime, 5960354062892656328

'ctime' means the time when the sheepdog cluster is created.  Did you
run 'collie cluster format' on the two machines at the different time?
You can see the more detailed info with the following commands:

 $ collie cluster info -a 172.29.22.74
 $ collie cluster info -a 172.29.22.78

If you clean the directory of either of the two nodes and restart the
sheep daemons, I think you can start Sheepdog.

> 
> but PC1 and PC2 use NTP and have exactly same date and time.

Sheepdog doesn't require the machines to have exactly same time.


Thanks,

Kazutaka


> Do you know why sheepdog can't start ?
> 
> 
> Thank you.
> 
> 
> 
> Note: my /etc/corosync/corosync.conf
> compatibility: whitetank
> totem {
>  version: 2
>  secauth: off
>  threads: 0
>  interface {
>  ringnumber: 0
>  bindnetaddr: 172.29.0.0
>  mcastaddr: 226.94.1.1
>  mcastport: 5405
>  }
>  }
>  logging {
>   fileline: off
>   to_stderr: no
>   to_logfile: yes
>   to_syslog: yes
>   logfile: /var/log/corosync.log
>   debug: off
>   timestamp: on
>   logger_subsys {
>    subsys: AMF
>    debug: off
>   }
>  }
>  amf {
>   mode: disabled
>  }
>