On Wed, Oct 20, 2010 at 10:41 AM, Colin Simpson <span dir="ltr">&lt;<a href="mailto:Colin.Simpson@iongeo.com" target="_blank">Colin.Simpson@iongeo.com</a>&gt;</span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">


Oct 20 15:47:44 testnode2 kernel: &quot;echo 0<br>
&gt; /proc/sys/kernel/hung_task_timeout_secs&quot; disables this message.<br></blockquote><div><br></div><div>That timeout is a warning from khungtaskd, it&#39;s not actually an error.  Just let&#39;s you know a kernel-thread was blocked for 120-seconds, but that may be OK, some tasks can take awhile to complete.  I&#39;ve only seen those when I do software raid resyncs.  If you didn&#39;t unmount the volume cleaning, it could be doing some recovery which is why you get the long-running thread.</div>
<div><br></div><div>I don&#39;t know what&#39;s going on with your setup.  I&#39;ve never had that issue, it could be something else in your configuration, corrupt data, etc.  It&#39;s not clear this is a DRBD issue, it could be cluster or GFS configuration.</div>
<div><br></div><div>I have rgmanager control all my GFS2 resources, example cluster.conf bits:</div><div><br></div><div><div>                &lt;resources&gt;</div><div>                        &lt;script file=&quot;/etc/init.d/httpd&quot; name=&quot;httpd&quot;/&gt;</div>
<div>                        &lt;script file=&quot;/etc/init.d/gfs2&quot; name=&quot;GFS2&quot;/&gt;</div><div>                        &lt;clusterfs device=&quot;/dev/ClusteredVG/gfs-vol&quot; force_unmount=&quot;0&quot; fsid=&quot;12345&quot; fstype=&quot;gfs2&quot; mountpoint=&quot;/content&quot; name=&quot;/content&quot; options=&quot;rw,noatime,nodiratime,noquota&quot; self_fence=&quot;0&quot;/&gt;</div>
<div>                &lt;/resources&gt;</div></div><div>                &lt;service autostart=&quot;1&quot; domain=&quot;failover1&quot; name=&quot;content&quot; recovery=&quot;restart&quot;&gt;</div><div><div>                        &lt;script ref=&quot;GFS2&quot;&gt;</div>
<div>                                &lt;clusterfs fstype=&quot;gfs&quot; ref=&quot;/content&quot;&gt;</div><div>                                        &lt;script ref=&quot;httpd&quot;/&gt;</div><div>                                &lt;/clusterfs&gt;</div>
<div>                        &lt;/script&gt;</div><div>                &lt;/service&gt;</div><div><br></div></div><div>The file system is also in fstab with &quot;noauto&quot; set for clean dismounting when rgmanager is shutdown.  It could be that you&#39;re not cleanly shutting down and recovering on startup for &gt;120 seconds.</div>
<div><br></div><div>-JR</div>

</div>