<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">On 01/01/2018 10:10 AM, Yaniv Kaul
      wrote:<br>
    </div>
    <blockquote type="cite"
cite="mid:CAJgorsY0bkoyp3fM2e0noMJQ-6W=_v1K-CP6b+SfrZr8o=qYhQ@mail.gmail.com">
      <div dir="ltr"><br>
        <div class="gmail_extra">
          <div class="gmail_quote">On Mon, Jan 1, 2018 at 12:50 AM,
            Andrei V <span dir="ltr">&lt;<a
                href="mailto:andreil1@starlett.lv" target="_blank"
                moz-do-not-send="true">andreil1@starlett.lv</a>&gt;</span>
            wrote:<br>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi !<br>
              <br>
              I'm installing 2-node failover cluster (2 x Xeon servers
              with local RAID<br>
              5 / ext4 for oVirt storage domains).<br>
              Now I have a dilemma - use either GlusterFS replica 2 or
              stick with NFS?<br>
            </blockquote>
            <div><br>
            </div>
            <div>Replica 2 is not good enough, as it can leave you with
              split brain. It's been discussed in the mailing list
              several times.</div>
            <div>How do you plan to achieve HA with NFS? With drbd?</div>
          </div>
        </div>
      </div>
    </blockquote>
    Hi, Yaniv,<br>
    Thanks a lot for detailed explanation!<br>
    <br>
    I know Replica 2 is not optimal solution. <br>
    Right now I have only 2 servers with internal RAIDs for nodes, and
    till end of this week system had to be running in whatever
    condition.<br>
    May be its better to use local storage domain on each node, set
    export domain on backup node, and backup VMs to 2nd backup node in
    timed interval? <br>
    Its not highly-available yet workable solution.<br>
    <br>
    <blockquote type="cite"
cite="mid:CAJgorsY0bkoyp3fM2e0noMJQ-6W=_v1K-CP6b+SfrZr8o=qYhQ@mail.gmail.com">
      <div dir="ltr">
        <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">
              4.2 Engine is running on separate hardware.<br>
            </blockquote>
            <div><br>
            </div>
            <div>Is the Engine also highly available?</div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    Its KVM appliance, could be launched on 2 SuSE servers.<br>
    <br>
    <blockquote type="cite"
cite="mid:CAJgorsY0bkoyp3fM2e0noMJQ-6W=_v1K-CP6b+SfrZr8o=qYhQ@mail.gmail.com">
      <div dir="ltr">
        <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">
              Each node have its own storage domain (on internal RAID).<br>
            </blockquote>
            <div><br>
            </div>
            <div>So some sort of replica 1 with geo-replication between
              them?</div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    Could it be the following?<br>
    1) Local storage domain on each node<br>
    2) GlusterFS geo-replication or over these directories? Not sure
    this will work.<br>
    <br>
    <blockquote type="cite"
cite="mid:CAJgorsY0bkoyp3fM2e0noMJQ-6W=_v1K-CP6b+SfrZr8o=qYhQ@mail.gmail.com">
      <div dir="ltr">
        <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">
              <br>
              All VMs must be highly available.<br>
            </blockquote>
            <div><br>
            </div>
            <div>Without shared storage, it may be tricky.</div>
          </div>
        </div>
      </div>
    </blockquote>
    <br>
    Seems to be timely VM backup to 2nd node is enough for this time.<br>
    With current hardware anything above is too cumbersome to setup.<br>
    <br>
    <blockquote type="cite"
cite="mid:CAJgorsY0bkoyp3fM2e0noMJQ-6W=_v1K-CP6b+SfrZr8o=qYhQ@mail.gmail.com">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">
            <div><br>
            </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              One of the VMs is an accounting/stock control system with
              FireBird SQL<br>
              server on CentOS is speed-critical.<br>
            </blockquote>
            <div><br>
            </div>
            <div>But is IO the bottleneck? Are you using SSDs / NVMe
              drives? </div>
            <div>I'm not familiar enough with FireBird SQL server - does
              it have an application layer replication you might opt to
              use?</div>
            <div>In such case, you could pass-through a NVM disk and
              have the application layer perform the replication between
              the nodes.</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              No load balancing between nodes necessary. 2nd is just for
              backup if 1st<br>
              for whatever reason goes up in smoke. All VM disks must be
              replicated to<br>
              backup node in near real-time or in worst case each 1 - 2
              hour.<br>
              GlusterFS solves this issue yet at high performance
              penalty.<br>
            </blockquote>
            <div><br>
            </div>
            <div>The problem with a passive backup is that you never
              know it'll really work when needed. This is why
              active-active is many time preferred.</div>
            <div>It's also more cost effective usually - instead of some
              HW lying around.</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              <br>
              &gt;From what I read here<br>
              <a
                href="http://lists.ovirt.org/pipermail/users/2017-July/083144.html"
                rel="noreferrer" target="_blank" moz-do-not-send="true">http://lists.ovirt.org/<wbr>pipermail/users/2017-July/<wbr>083144.html</a><br>
              GlusterFS performance with oVirt is not very good right
              now because QEMU<br>
              uses FUSE instead of libgfapi.<br>
              <br>
              What is optimal way to go on ?<br>
            </blockquote>
            <div><br>
            </div>
            <div>It's hard to answer without additional details.</div>
            <div>Y.</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              Thanks in advance.<br>
              Andrei<br>
              <br>
              ______________________________<wbr>_________________<br>
              Users mailing list<br>
              <a href="mailto:Users@ovirt.org" moz-do-not-send="true">Users@ovirt.org</a><br>
              <a href="http://lists.ovirt.org/mailman/listinfo/users"
                rel="noreferrer" target="_blank" moz-do-not-send="true">http://lists.ovirt.org/<wbr>mailman/listinfo/users</a><br>
            </blockquote>
          </div>
          <br>
        </div>
      </div>
    </blockquote>
    <p><br>
    </p>
  </body>
</html>