[Gluster-users] When self-healing is triggered?

Rodrigo Severo rodrigo at fabricadeideias.com
Tue May 29 14:21:45 UTC 2012


On Tue, May 29, 2012 at 11:05 AM, Flavio Oliveira <nvezes at live.com> wrote:

>  Hi,
>
> I am new in the Gluster world but I got a problem in my hands.
>
> Actually, as the client was disconnected and re-connected to the server,
> it may be enough reason to trigger the self-healing. That was my first
> thought.
> The servers may not be synchronized (Timeout, Pending actions, etc).
> However a more experient co-worker told me that it was not enough reason.
> Anyway, what I am really investigating is the timeout message below, it
> might be connected with the self-healing. Maybe something went wrong during
> the
> self-healing (File system corrupted!?). So, that's why I need to
> understand when the self-healing is triggered.
>
> [2012-05-22 21:57:09.627220] E [rpc-clnt.c:199:call_bail] 0-client-6:
> bailing out frame type(GlusterFS 3.1) op(FINODELK(30)) xid = 0x358492x sent
> = 2012-05-22 21:27:00.394792. timeout = 1800
>
> Basically we have two FTP servers that upload data to the storages and
> Glusterfs will write files synchronously on all replicas.
> So the files become available to On Demand streaming.
>
>
> Customer -> FTP Server -> Storage <-> Gluster Clienter <-> Viewer
>

It's getting clearer. Focusing on the Gluster infrastructures: there are
Gluster servers and Gluster clients. Where are them in the above schema?

Are the storage servers the gluster servers? You know you shall not change
anything directly in a brick, only through Gluster native client, Gluster
NFS or Gluster CIFS?


Regards,

Rodrigo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20120529/eecdf94f/attachment.html>


More information about the Gluster-users mailing list