<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">02.02.2014 13:53, Valerio Pachera
      пишет:</div>
    <blockquote
cite="mid:CAHS0cb_SiK30omOypjQGd=fF157tARdTaRkyionURqdn6M7VVQ@mail.gmail.com"
      type="cite">
      <div dir="ltr"><br>
        <div class="gmail_extra">2014-02-01 Maxim Terletskiy <span
            dir="ltr"><<a moz-do-not-send="true"
              href="mailto:terletskiy@emu.ru" target="_blank">terletskiy@emu.ru</a>></span>:<br>
          <div class="gmail_quote">
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              the reason before found that "nodiratime" is root of the
              evil. </blockquote>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">Hope
              this information would be usefull for somebody.<br>
            </blockquote>
            <div><br>
              Sure it  is.<br>
            </div>
            <div>Has any body else been able to reproduce this problem?<br>
            </div>
            <div>What version of sheep are you running?<br>
            </div>
            <div><br>
            </div>
          </div>
        </div>
      </div>
    </blockquote>
    For me "nodiratime" problem arises with QEMU 1.6.1, 1.6.2, 1.7.0.
    Using CentOS 6.5 x86_64, sheep version 0.7.6 with zookeeper, object
    cache on ext4 partition, dual-nic with bonded ethernet for cluster
    communication and IPoIB for IO. Sheep options I'm using on VM nodes:<br>
    sheep -g -c
    zookeeper:172.16.0.15:2181,172.31.200.249:2181,172.31.200.250:2181,timeout=30
    -y 172.16.0.10 -i host=192.168.10.10,port=7001 -w
    dir=/mnt/sheep_objcache,size=213000<br>
    <br>
    <blockquote
cite="mid:CAHS0cb_SiK30omOypjQGd=fF157tARdTaRkyionURqdn6M7VVQ@mail.gmail.com"
      type="cite">
      <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">
              Now I'm curious about object cache failover. What happen
              if volume with cache will fail? Will sheep and VMs live or
              will they die?<br>
            </blockquote>
            <div><br>
            </div>
            <div>I'm also curious.<br>
            </div>
            <div>I think it's not handled as of now.<br>
            </div>
            <div>It would be good that, in case of I/O error from the
              cache device, sheep disables cache (like a node md
              unplug).<br>
            </div>
          </div>
        </div>
      </div>
    </blockquote>
    Morita Kazukata writes that VMs must survive. I will try to do some
    tests.<br>
    <blockquote
cite="mid:CAHS0cb_SiK30omOypjQGd=fF157tARdTaRkyionURqdn6M7VVQ@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              (I found that with restart of sheep process VMs connected
              to it dying)<br>
            </blockquote>
            <div><br>
            </div>
            <div>They shouldn't, but this depends on your qemu version.<br>
            </div>
            <div>Qemu supports auto-reconnect. As soon as sheep daemon
              is available again, the vm will issue their I/O requests.<br>
            </div>
            <div> </div>
          </div>
        </div>
      </div>
    </blockquote>
    Tried twice with QEMU 1.6.1. Both times result was negative.<br>
  </body>
</html>