[Sheepdog] Configuring Sheepdog environment

[MORITA Kazutaka]

On 2011/06/22 4:24, Gustavo Callou wrote:
> Dear Kazutaka ,
> 
> Attached to this message you have the logs.
> Related to the command "collie cluster info", I added a screenshot to show
> what is happening in all cluster nodes.

Your sheep.log says that the sheep daemon cannot access to the epoch
directory (~/sheepStorage/epoch).
==
  Jun 20 11:40:24 read_epoch(2015) failed to read epoch 4294967295
==
Perhaps, you don't have a permission to access the directory?

I've pushed a patch (http://lists.wpkg.org/pipermail/sheepdog/2011-June/001105.html)
to show the error reason.
Could you try again with the latest snapshot in https://github.com/collie/sheepdog ?
I think the sheep daemon will abort just after you start it,
then please send me the output.

Thanks,

Kazutaka

> 
> Best regards,
> Gustavo
> 
> 
> On Tue, Jun 21, 2011 at 9:56 AM, MORITA Kazutaka <
> morita.kazutaka at lab.ntt.co.jp> wrote:
> 
>> On Tue, Jun 21, 2011 at 4:37 AM, Gustavo Callou <grac at cin.ufpe.br> wrote:
>>> Dear Sheepdog developers,
>>>
>>> First of all, we would like to thank you for the support.
>>>
>>> We have tried to install the sheepdog version that you provided for us.
>>> However, we were not successful with the installation due to the fact
>> that
>>> we could not obtain the result of the command collie node list (e.g.,
>>> something like connection failure). Thus, we adopted the sheepdog version
>>> available at https://github.com/collie/sheepdog
>>> (collie-sheepdog-v0.2.3-11-g3410623.tar.gz). Considering this version, we
>>> were successful to see all machines in the cluster (adopting the command
>>> collie node list). However, the command collie cluster format –copies=2
>>> seems to be not working as it is supposed to (e.g., Invalid error code) .
>>> Besides that, we tried the command collie node info , and the result was:
>>>
>>> Id Size Used Use%
>>>
>>> cannot get information from any nodes
>>>
>>> Another command “collie cluster info ” seems also to be not working very
>>> well (we could not see the creation time and epoch nodes)
>>>
>>> Cluster status: running
>>>
>>> Creation time Epoch Nodes
>>
>> Can you give me the results of "collie cluster info" on the other nodes?
>>
>>>
>>> In addition, although the command “qemu-img create sheepdog:Callou 4G”
>>> seemed to work, once it has executed the node is out of the cluster.
>> Besides
>>
>> Hmm, this command shouldn't work if the Sheepdog cluster is not
>> formatted...
>>
>>
>>> that, we executed the command “collie vdi list” , and we got:
>>>
>>> name id size used shared creation time vdi id
>>>
>>> ------------------------------------------------------------------
>>> failed to read a inode header 15507205, 0, 81
>>>
>>> Thus, we are not sure if the vm image was created. In order to put the
>> node
>>> again into the cluster, we had to restart the corosync and sheepdog as
>> well
>>> as to delete an create again the sheepdog storage dir.
>>>
>>> Then, we executed the command “collie vdi list” and got:
>>>
>>> name id size used shared creation time vdi id
>>>
>>> ------------------------------------------------------------------
>>>
>>> Floating point exception
>>>
>>> We expect that those information provided above may be useful to find out
>>> what is the problem that might be happening in my sheepdog environment.
>> Do
>>> you have any suggestion?
>>
>> Your previous log files showed that different versions of sheep
>> daemons coexist on your cluster.
>> Could you check that you are using the same versions of Sheepdog on
>> all the machines?
>>
>> If it still doesn't work, can you give me the sheep.log again?
>>
>> Thanks,
>>
>> Kazutaka
>>
> 
> 
>