<div dir="ltr"><br><div class="gmail_quote">On Fri, Aug 22, 2008 at 7:14 PM, Lars Ellenberg <span dir="ltr"><<a href="mailto:lars.ellenberg@linbit.com">lars.ellenberg@linbit.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="Ih2E3d">[snip]<br>
<br></div>right.<br>
nfsnode1 is not the home node,<br>
and apparently concludes from seeing the home node alive that the<br>
resources are running there alright, and takes no action.<br>
</blockquote><div> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">[snip]<br>
</blockquote><div> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">aparently heartbeat in haresources (non-crm) mode<br>
cannot cope with your scenario.<br>
<div class="Ih2E3d">[snip]<br>
<br>
</div>if you want to cope with multiple failures, operator intervention is<br>
almost always required. in this scenario, if you want to stick with<br>
haresources style heartbeat config, you probably have to say something like<br>
# /usr/lib/heartbeat/ResourceManager takegroup drbddisk::drbd-resource-0<br>
could do the trick now. (but don't tell anybody)<br>
<br>
it may even be a heartbeat "bug", but I doubt that linux-ha guys are<br>
eager to change anything in the officially unmaintained "non-crm code".<br>
<br>
</blockquote></div><br>Ok, thanks for your clarifications.<br>The command provided makes the right work (I didn't tell anybody ;-).<br>I started with 1.x style config as all the documents about drbd+heartbeat+nfs are based on this.<br>
I saw that one of main features of 2.x heartbeat should be resource monitoring, being able to cope with node failures, network failures such as full link failures or the cluster becoming partitioned, but also with failures of individual resources,<br>
so that in my second step I'm going to migrate to this config.<br><br>BTW, I was also testing the primary-poweroff scenario and it didn't work.<br>But this was caused by a severe bug in 2.1.3 heartbeat dopd component, as released at this time in CentOS 5.2.<br>
See <br><a href="http://bugs.centos.org/view.php?id=2875">http://bugs.centos.org/view.php?id=2875</a><br><br><a href="http://hg.linux-ha.org/dev/rev/47f60bebe7b2">http://hg.linux-ha.org/dev/rev/47f60bebe7b2</a><br><br>So I needed to recompile heartbeat.<br>
But it was not related with the scenario posted here, where all goes as you clearly described.<br><br>In case anyone needs too....<br>while I'm writing, I see that 2.1.4 has just been released ...... fixing the problems...<br>
so I have also to see changes between 2.1.3 and <a href="http://2.1.4.">2.1.4.</a>. '-)<br><br>Thanks again.<br>Gianluca<br></div>