Re: 3.3.0 -> 3.4.2 / Rolling upgrades with no downtime

Bryan Whitehead <driver@xxxxxxxxxxxxxxxxxxxxxx> · Wed, 5 Mar 2014 12:26:03 -0800

I followed this blog: http://geekshell.org/~pivo/glusterfs-split-brain.html
(this can take too long because of using "find" if you have many files)

Right after Joe Julian released a pretty handy system for exploring the split brains and fixing. You can check it out here:
http://joejulian.name/blog/glusterfs-split-brain-recovery-made-easy/

https://github.com/joejulian/glusterfs-splitbrain

On Wed, Mar 5, 2014 at 1:40 AM, João Pagaime <joao.pagaime@xxxxxxxxx> wrote:

    hello Bryan and thanks for sharing!

      how did you fix those 2 files on a  split-brain situation? deleted
      one "bad" file?

           which one to select for
            deletion?

            on an software update situation I would expect not to have
            peer probe problems, simply because there are no "gluster
            peer probe" commands. The problem that could happen is the
            updated 3.4.2 node having problems reentering a  3.3.0
            cluster (without "peer probes" commands). It's good news it
            went well

            best regards

            joao

      Em 04-03-2014 18:30, Bryan Whitehead escreveu:

        I just did this last week from 3.3.0->3.4.2.

          I never got the peer probe problems - but I did end up with 2
          files being in a split-brain situation.

        Note: I only had ~hundred files that are qcow2 for KVM, so 2
        files getting split-brain is about 2% filesystem problem.

        On Tue, Mar 4, 2014 at 1:43 AM, João
          Pagaime <joao.pagaime@xxxxxxxxx>
          wrote:

          Hello all

            anyone tried a  rolling upgrades with no downtime [1] from
            3.3.0 to 3.4.2 or similar upgrade? any comments?

            for testing purposes we've installed a 3.4.2 server and it
            won't peer, giving the error "peer probe: failed: Peer X
            does not support required op-version".

            I guess this is expected behavior for a new entry on the
            cluster

            What about changing the software on an existing peer of the
            cluster? Will it also refuse to re-enter the cluster after
            the upgrade for the same reason (peers not supporting the
            required op-version)?

            After all servers and clients are upgraded, how to increase
            the op-version of the global cluster?

            best regards,

            joão

            [1]

            http://vbellur.wordpress.com/2013/07/15/upgrading-to-glusterfs-3-4/

            _______________________________________________

            Gluster-users mailing list

            Gluster-users@xxxxxxxxxxx

            http://supercolony.gluster.org/mailman/listinfo/gluster-users

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://supercolony.gluster.org/mailman/listinfo/gluster-users