[Gluster-users] Gluster native client mount not accessible temporarily when one of the replica restart

[Jestan Nirojan]

Hi all,

I am setting up a gluster replica with 2 bricks and 2 native client mounts
on same the bricks servers.


In server1 :
                  volume "logs" brick 1  on /data/brick1 (XFS)

                  mount  server1:/logs on /mount/logs

In server2:

                  volume "logs" brick 2  on /data/brick2 (XFS)

                  mount  server2:/logs on /mount/logs


gluster version 3.4.6

When I reboot one of the brick servers, other brick server's mount is not
accessible temporarily, after few seconds, it becomes available for
reading/writing.
Any ideas on how to avoid this issue? (reducing network.ping-timeout to 1
second give better results, but still the issue there for 1 second)

I have an application1: writing logs in sever1 /mount/logs and
               application2: reading logs in sever2 /mount/logs

write/read performance is good in sever1 /mount/logs , read is very slow in
server2  /mount/logs.

but reading the files directly in server2 is fast compare to reading it
through the mount point, why is that?

is there any way to configure the native client to read from local brick?
(setting cluster.choose-local : on, didn't improve the read performance)


thanks and regards,
- Jestan Nirojan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150218/568e03ec/attachment.html>