<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">On 09/29/2016 05:18 PM, Sahina Bose
      wrote:<br>
    </div>
    <blockquote
cite="mid:CACjzOvfCdjG0H+pi7hf_jVNxhJfv-Qi2r3u-yTqnYGPt1W0O2w@mail.gmail.com"
      type="cite">
      <div dir="ltr">Yes, this is a GlusterFS problem. Adding gluster
        users ML<br>
      </div>
      <div class="gmail_extra"><br>
        <div class="gmail_quote">On Thu, Sep 29, 2016 at 5:11 PM, Davide
          Ferrari <span dir="ltr">&lt;<a moz-do-not-send="true"
              href="mailto:davide@billymob.com" target="_blank">davide@billymob.com</a>&gt;</span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div dir="ltr">
              <div>
                <div>
                  <div>
                    <div>
                      <div>
                        <div>Hello<br>
                          <br>
                        </div>
                        maybe this is more glustefs then ovirt related
                        but since OVirt integrates Gluster management
                        and I'm experiencing the problem in an ovirt
                        cluster, I'm writing here.<br>
                        <br>
                      </div>
                      The problem is simple: I have a data domain
                      mappend on a replica 3 arbiter1 Gluster volume
                      with 6 bricks, like this:<br>
                      <span style="font-family:monospace,monospace"><br>
                        Status of volume: data_ssd<br>
                        Gluster process                       <wbr>     
                        TCP Port  RDMA Port  Online  Pid<br>
                        ------------------------------<wbr>------------------------------<wbr>------------------<br>
                        Brick vm01.storage.billy:/gluster/<wbr>ssd/data/<br>
                        brick                         <wbr>             
                        49153     0          Y       19298<br>
                        Brick vm02.storage.billy:/gluster/<wbr>ssd/data/<br>
                        brick                         <wbr>             
                        49153     0          Y       6146 <br>
                        Brick vm03.storage.billy:/gluster/<wbr>ssd/data/<br>
                        arbiter_brick                 <wbr>             
                        49153     0          Y       6552 <br>
                        Brick vm03.storage.billy:/gluster/<wbr>ssd/data/<br>
                        brick                         <wbr>             
                        49154     0          Y       6559 <br>
                        Brick vm04.storage.billy:/gluster/<wbr>ssd/data/<br>
                        brick                         <wbr>             
                        49152     0          Y       6077 <br>
                        Brick vm02.storage.billy:/gluster/<wbr>ssd/data/<br>
                        arbiter_brick                 <wbr>             
                        49154     0          Y       6153 <br>
                        Self-heal Daemon on localhost              
                        N/A       N/A        Y       30746<br>
                        Self-heal Daemon on vm01.storage.billy     
                        N/A       N/A        Y       196058<br>
                        Self-heal Daemon on vm03.storage.billy     
                        N/A       N/A        Y       23205<br>
                        Self-heal Daemon on vm04.storage.billy     
                        N/A       N/A        Y       8246 </span><br>
                      <br>
                      <br>
                    </div>
                    Now, I've put in maintenance the vm04 host, from
                    ovirt, ticking the "Stop gluster" checkbox, and
                    Ovirt didn't complain about anything. But when I
                    tried to run a new VM it complained about "storage
                    I/O problem", while the storage data status was
                    always UP.<br>
                    <br>
                  </div>
                  Looking in the gluster logs I can see this:<br>
                  <br>
                  <span style="font-family:monospace,monospace">[2016-09-29
                    11:01:01.556908] I [glusterfsd-mgmt.c:1596:mgmt_<wbr>getspec_cbk]
                    0-glusterfs: No change in volfile, continuing<br>
                    [2016-09-29 11:02:28.124151] E [MSGID: 108008]
                    [afr-read-txn.c:89:afr_read_<wbr>txn_refresh_done]
                    0-data_ssd-replicate-1: Failing READ on gfid
                    bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d:
                    split-brain observed. [Input/output error]<br>
                    [2016-09-29 11:02:28.126580] W [MSGID: 108008]
                    [afr-read-txn.c:244:afr_read_<wbr>txn]
                    0-data_ssd-replicate-1: Unreadable subvolume -1
                    found with event generation 6 for gfid
                    bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d. (Possible
                    split-brain)<br>
                    [2016-09-29 11:02:28.127374] E [MSGID: 108008]
                    [afr-read-txn.c:89:afr_read_<wbr>txn_refresh_done]
                    0-data_ssd-replicate-1: Failing FGETXATTR on gfid
                    bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d:
                    split-brain observed. [Input/output error]<br>
                    [2016-09-29 11:02:28.128130] W [MSGID: 108027]
                    [afr-common.c:2403:afr_<wbr>discover_done]
                    0-data_ssd-replicate-1: no read subvols for (null)<br>
                    [2016-09-29 11:02:28.129890] W
                    [fuse-bridge.c:2228:fuse_<wbr>readv_cbk]
                    0-glusterfs-fuse: 8201: READ =&gt; -1
                    gfid=bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d
                    fd=0x7f09b749d210 (Input/output error)<br>
                    [2016-09-29 11:02:28.130824] E [MSGID: 108008]
                    [afr-read-txn.c:89:afr_read_<wbr>txn_refresh_done]
                    0-data_ssd-replicate-1: Failing FSTAT on gfid
                    bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d:
                    split-brain observed. [Input/output error]<br>
                  </span></div>
              </div>
            </div>
          </blockquote>
        </div>
      </div>
    </blockquote>
    <br>
    Does `gluster volume heal data_ssd info split-brain` report that the
    file is in split-brain, with vm04 still being down? <br>
    If yes, could you provide the extended attributes of this gfid from
    all 3 bricks:<br>
    getfattr -d -m . -e hex /path/to/brick/bf/59/<span
      style="font-family:monospace,monospace">bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d</span><br>
    <br>
    If no, then I'm guessing that it is not in actual split-brain (hence
    the 'Possible split-brain' message). If the node you brought down
    contains the only good copy of the file (i.e the other data brick
    and arbiter are up, and the arbiter 'blames' this other brick), all
    I/O is failed with EIO to prevent file from getting into actual
    split-brain. The heals will happen when the good node comes up and
    I/O should be allowed again in that case.<br>
    <br>
    -Ravi<br>
    <br>
    <br>
    <blockquote
cite="mid:CACjzOvfCdjG0H+pi7hf_jVNxhJfv-Qi2r3u-yTqnYGPt1W0O2w@mail.gmail.com"
      type="cite">
      <div class="gmail_extra">
        <div class="gmail_quote">
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div dir="ltr">
              <div>
                <div><span style="font-family:monospace,monospace">[2016-09-29
                    11:02:28.133879] W [fuse-bridge.c:767:fuse_attr_<wbr>cbk]
                    0-glusterfs-fuse: 8202: FSTAT()
                    /ba2bd397-9222-424d-aecc-<wbr>eb652c0169d9/images/f02ac1ce-<wbr>52cd-4b81-8b29-f8006d0469e0/<wbr>ff4e49c6-3084-4234-80a1-<wbr>18a67615c527
                    =&gt; -1 (Input/output error)<br>
                    The message "W [MSGID: 108008]
                    [afr-read-txn.c:244:afr_read_<wbr>txn]
                    0-data_ssd-replicate-1: Unreadable subvolume -1
                    found with event generation 6 for gfid
                    bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d. (Possible
                    split-brain)" repeated 11 times between [2016-09-29
                    11:02:28.126580] and [2016-09-29 11:02:28.517744]<br>
                    [2016-09-29 11:02:28.518607] E [MSGID: 108008]
                    [afr-read-txn.c:89:afr_read_<wbr>txn_refresh_done]
                    0-data_ssd-replicate-1: Failing STAT on gfid
                    bf5922b7-19f3-4ce3-98df-<wbr>71e981ecca8d:
                    split-brain observed. [Input/output error]<br>
                  </span><br>
                </div>
                Now, how is it possible to have a split brain if I
                stopped just ONE server which had just ONE of six
                bricks, and it was cleanly shut down with maintenance
                mode from ovirt?<br>
                <br>
              </div>
              I created the volume originally this way:<br>
              <span style="font-family:monospace,monospace"># gluster
                volume create data_ssd replica 3 arbiter 1
                vm01.storage.billy:/gluster/<wbr>ssd/data/brick
                vm02.storage.billy:/gluster/<wbr>ssd/data/brick
                vm03.storage.billy:/gluster/<wbr>ssd/data/arbiter_brick
                vm03.storage.billy:/gluster/<wbr>ssd/data/brick
                vm04.storage.billy:/gluster/<wbr>ssd/data/brick
                vm02.storage.billy:/gluster/<wbr>ssd/data/arbiter_brick<br>
                # gluster volume set data_ssd group virt<br>
                # gluster volume set data_ssd storage.owner-uid 36
                &amp;&amp; gluster volume set data_ssd storage.owner-gid
                36<br>
                # gluster volume start data_ssd</span><span
                class="HOEnZb"><font color="#888888"><br>
                </font></span></div>
          </blockquote>
        </div>
      </div>
    </blockquote>
    <br>
    <br>
    <br>
    <br>
    <blockquote
cite="mid:CACjzOvfCdjG0H+pi7hf_jVNxhJfv-Qi2r3u-yTqnYGPt1W0O2w@mail.gmail.com"
      type="cite">
      <div class="gmail_extra">
        <div class="gmail_quote">
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div dir="ltr"><span class="HOEnZb"><font color="#888888"><br
                    clear="all">
                  <div>
                    <div>
                      <div>
                        <div>
                          <div>
                            <div>
                              <div><br>
                                -- <br>
                                <div>
                                  <div dir="ltr">
                                    <div>Davide Ferrari<br>
                                    </div>
                                    Senior Systems Engineer<br>
                                  </div>
                                </div>
                              </div>
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </font></span></div>
            <br>
            ______________________________<wbr>_________________<br>
            Users mailing list<br>
            <a moz-do-not-send="true" href="mailto:Users@ovirt.org">Users@ovirt.org</a><br>
            <a moz-do-not-send="true"
              href="http://lists.ovirt.org/mailman/listinfo/users"
              rel="noreferrer" target="_blank">http://lists.ovirt.org/<wbr>mailman/listinfo/users</a><br>
            <br>
          </blockquote>
        </div>
        <br>
      </div>
    </blockquote>
    <p><br>
    </p>
  </body>
</html>