<div dir="ltr"><br><div class="gmail_extra"><br><br><div class="gmail_quote">2014-06-26 17:01 GMT+02:00 Valerio Pachera <span dir="ltr"><<a href="mailto:sirio81@gmail.com" target="_blank">sirio81@gmail.com</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div><div><div>I disconnected a node by dog node kill.<br></div>After recovery I stopped and restarted the cluster because I had to move some cables.<br>
</div>After that, I run one of the two guest but It was comlaining about filesystem errors.<br></div></div></blockquote><br></div><div class="gmail_quote">After a second thought, I didn't check the guest status before disconnecting the node (before recovery).<br>
</div><div class="gmail_quote">The node I disconnected in past 2 days was working slowly because of the raid hardware controller cache not enabled.<br></div><div class="gmail_quote">I'm almost sure the corruption happened because of that.<br>
</div><div class="gmail_quote">The guest receives data from another server.<br>In the server's backup logs of the last night I read many of these errors:<br><br>rsync: recv_generator: mkdir "/opt/zimbra/data/amavisd/tmp/amavis-20140625T092245-03629" failed: Read-only file system.<br>
<br></div><div class="gmail_quote">Andrew J.H. said sheep daemon v0.7 was crashing in similar circumstances.<br></div><div class="gmail_quote">It seems 0.82 doesn't crash but doesn't handle it well.<br><br></div>
<div class="gmail_quote">Did any of you experienced that?<br><br></div><div class="gmail_quote">This may be replicated forcing the link of a nic at 10M/s<br><code>ethtool -s eth0 speed 10 duplex full<br></code></div><div class="gmail_quote">
<br>Maybe even running sheep without '-n' option (only on one of the nodes).<br></div></div></div>