[Gluster-devel] [Gluster-users] Peers rejected, broken setup

[Nux!]

On 26.09.2013 03:41, Vijay Bellur wrote:
> On 09/26/2013 12:45 AM, Nux! wrote:
>> On 25.09.2013 18:24, Vijay Bellur wrote:
>>> On 09/25/2013 08:50 PM, Nux! wrote:
>>>> On 25.09.2013 15:42, Nux! wrote:
>>>>> Hello,
>>>>> 
>>>>> I'm currently using 4 servers with various types of volumes on 
>>>>> them,
>>>>> today I noticed 2 of them report the other 2 as "peer rejected" 
>>>>> and
>>>>> viceversa.
>>>>> Where do I even begin to debug this? I don't see anything 
>>>>> meaningful
>>>>> in the logs.
>>>>> Any pointers?
>>> 
>>> "Peer rejected" is seen if there is a volume configuration mismatch
>>> between peers. Can you check if there is a mismatch in volume
>>> configuration between the two sets of peers?
>> 
>> I have checked the volumes "info" files following Krishnan's advice 
>> and
>> they seem to have the same data, though oddly not exactly in the same
>> place, e.g. some have auth.allow above rpc.auth-allow and it's
>> counterpart info files from other servers have it the other way 
>> around,
>> but the values are identical. Aren't these files supposed to be 100% 
>> the
>> same?
> 
> The ordering need not be the same. Can you please verify if there's a
> difference in the content of this file on all peers:
> 
> /var/lib/glusterd/vols/<volname>/cksum

That file seems to differ between pair of servers, so it's the same on 
the replica members, but it's different from the one on the rejected 
peers.
Only one volume has the same cksum the same across all servers.

What's next?

-- 
Sent from the Delta quadrant using Borg technology!

Nux!
www.nux.ro