[Gluster-devel] Brick is Offline
ABHISHEK PALIWAL
abhishpaliwal at gmail.com
Fri Apr 1 10:53:18 UTC 2016
Please check the cmd-history file in attached logs of the Board A in which
we are removing-brick at 11:56:48 because at this time after waiting for 1
minute brick was not online.
So, We need to identify near to this time why brick is not online.
Because If brick will be online we will not remove-brick and if we remove
brick, volume become distributed which will also cause problem in our case.
Regards,
Abhishek
On Fri, Apr 1, 2016 at 4:08 PM, Atin Mukherjee <amukherj at redhat.com> wrote:
>
>
> On 04/01/2016 03:57 PM, ABHISHEK PALIWAL wrote:
> >
> >
> > On Fri, Apr 1, 2016 at 3:39 PM, Atin Mukherjee <amukherj at redhat.com
> > <mailto:amukherj at redhat.com>> wrote:
> >
> >
> >
> > On 04/01/2016 03:06 PM, ABHISHEK PALIWAL wrote:
> > >
> > > On Fri, Apr 1, 2016 at 2:59 PM, Atin Mukherjee <
> amukherj at redhat.com <mailto:amukherj at redhat.com>
> > > <mailto:amukherj at redhat.com <mailto:amukherj at redhat.com>>> wrote:
> > >
> > >
> > >
> > > On 04/01/2016 02:55 PM, ABHISHEK PALIWAL wrote:
> > > > Hi Atin,
> > > >
> > > > Thanks for reply.
> > > >
> > > > Could you please help me to identify the error log in the
> respective
> > > > brick log file. I tried but not able to identified where the
> problem is
> > > > occuring.
> > > >
> > > > I am attaching the brick log file which is not coming online
> even after
> > > > waiting for 1 minute.
> > > What time did you reboot B? Could you also attach glusterd log
> file
> > > (complete log) for board B?
> > > >
> > >
> > > it is hard to say at what time we rebooted the B board because we
> are
> > > continuously rebooting the board B.
> > >
> > > Here I am attaching the glusterd and glsuterfs log for board B.
> > I can see the last restart of glusterd was at 13:11:27 and as per the
> > brick log there was no restart. This doesn't look like a reboot of
> the
> > board as in that case brick process should have also died and
> restarted.
> > Brick log indicates that the process is still running and there is no
> > interruption.
> >
> > In brick log file at 13:11:27 we have the logs showing reboot of board
> >
> > [2016-03-31 13:11:27.408512] I [MSGID: 115036]
> > [server.c:552:server_rpc_notify] 0-c_glusterfs-server: disconnecting
> > connection from
> > 002500-10939-2016/03/31-11:56:52:528771-c_glusterfs-client-11-0-0
> > [2016-03-31 13:11:27.408603] I [MSGID: 101055]
> > [client_t.c:419:gf_client_unref] 0-c_glusterfs-server: Shutting down
> > connection
> > 002500-10939-2016/03/31-11:56:52:528771-c_glusterfs-client-11-0-0
> FYI..this log indicates client disconnection.
> >
> > >
> > > > Regards,
> > > > Abhishek
> > > >
> > > > On Fri, Apr 1, 2016 at 1:04 PM, Atin Mukherjee <
> amukherj at redhat.com <mailto:amukherj at redhat.com>
> > <mailto:amukherj at redhat.com <mailto:amukherj at redhat.com>>
> > > > <mailto:amukherj at redhat.com <mailto:amukherj at redhat.com>
> > <mailto:amukherj at redhat.com <mailto:amukherj at redhat.com>>>> wrote:
> > > >
> > > >
> > > >
> > > > On 04/01/2016 12:10 PM, ABHISHEK PALIWAL wrote:
> > > > > Hi,
> > > > >
> > > > >
> > > > > I have the setup of two boards A and B with two bricks
> in
> > > replica mode.
> > > > >
> > > > > There is one test scenario
> > > > >
> > > > > 1. A acts as an active board and having the glusterfs
> > mount
> > > point on it.
> > > > > 2. B acts as Passive board.
> > > > > 3. We are repetitively rebooting the B board (In this
> time
> > > period peer
> > > > > status on A board will be "peer in cluster
> (Disconnected)"
> > > and brick is
> > > > > not present in "gluster volume status") and when Board
> B
> > > comes up,
> > > > > starts the gluster daemon.
> > > > > 4. if Gluster daemon starts successfully it will make
> > "peer in
> > > > > cluster(Connected)"
> > > > > 5. At the same with the immediate effect "gluster
> volume
> > > status" command
> > > > > should show the brick is available in online.
> > > > >
> > > > >
> > > > > But in my case sometime step 5 takes immediate
> reflection
> > > sometime 10-15
> > > > > second and sometime doesn't show brick is online even
> > after
> > > the 1minute.
> > > > >
> > > > > Could you please confirm why this type of unpredictable
> > > behavior is
> > > > > occuring. It should be reflect with immediate effect in
> > > "gluster volume
> > > > > status" command.
> > > > When glusterd restarts bricks processes are brought up
> > > asynchronously
> > > > and hence you may not see the brick processes reflecting
> in
> > > gluster
> > > > volume status output immediately after restart. 5-10
> > seconds is an
> > > > accepted time frame. However if it doesn't come back
> online
> > > post that
> > > > then probably brick fails to start in that case.
> > > >
> > > > Please check the respective brick log file and if you
> > can find
> > > any error
> > > > logs in it.
> > > > >
> > > > >
> > > > > --
> > > > > Regards
> > > > > Abhishek Paliwal
> > > > >
> > > > >
> > > > > _______________________________________________
> > > > > Gluster-devel mailing list
> > > > > Gluster-devel at gluster.org
> > <mailto:Gluster-devel at gluster.org> <mailto:Gluster-devel at gluster.org
> > <mailto:Gluster-devel at gluster.org>>
> > > <mailto:Gluster-devel at gluster.org
> > <mailto:Gluster-devel at gluster.org> <mailto:Gluster-devel at gluster.org
> > <mailto:Gluster-devel at gluster.org>>>
> > > > > http://www.gluster.org/mailman/listinfo/gluster-devel
> > > > >
> > > >
> > > >
> > > >
> > > >
> > > > --
> > > >
> > > >
> > > >
> > > >
> > > > Regards
> > > > Abhishek Paliwal
> > >
> > >
> >
> >
> >
> >
> > --
> >
> >
> >
> >
> > Regards
> > Abhishek Paliwal
>
--
Regards
Abhishek Paliwal
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160401/839d54fd/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Board A logs.tar.gz
Type: application/x-gzip
Size: 45433 bytes
Desc: not available
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160401/839d54fd/attachment-0001.gz>
More information about the Gluster-devel
mailing list