<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">2015-06-16 19:30 GMT+02:00 Digimer <span dir="ltr">&lt;<a href="mailto:lists@alteeve.ca" target="_blank">lists@alteeve.ca</a>&gt;</span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On 16/06/15 04:18 AM, Oscar Salvador wrote:<br>
&gt;<br>
&gt;<br>
&gt; 2015-06-16 5:59 GMT+02:00 Andrew Beekhof &lt;<a href="mailto:andrew@beekhof.net">andrew@beekhof.net</a><br>
</span>&gt; &lt;mailto:<a href="mailto:andrew@beekhof.net">andrew@beekhof.net</a>&gt;&gt;:<br>
<span class="">&gt;<br>
&gt;<br>
&gt;     &gt; On 16 Jun 2015, at 12:00 am, Oscar Salvador &lt;<a href="mailto:osalvador.vilardaga@gmail.com">osalvador.vilardaga@gmail.com</a><br>
</span><span class="">&gt;     &lt;mailto:<a href="mailto:osalvador.vilardaga@gmail.com">osalvador.vilardaga@gmail.com</a>&gt;&gt; wrote:<br>
&gt;     &gt;<br>
&gt;     &gt; Hi,<br>
&gt;     &gt;<br>
&gt;     &gt; I&#39;ve configured a fencing with libvirt, but I&#39;m having some<br>
&gt;     problem with stonith, due to the error &quot;no route to host”<br>
&gt;<br>
&gt;     That message is a bit wonky.<br>
&gt;     What it really means is that there were no devices that advertise<br>
&gt;     the ability to fence that node.<br>
&gt;<br>
&gt;     In this case, pacemaker wants to fence “server” but hostlist is set<br>
&gt;     to server.fqdn<br>
&gt;     Drop the .fqdn and it should work<br>
&gt;<br>
&gt;<br>
&gt; Get rid of the +fqdn was not an option, sorry, but I could fix it in<br>
&gt; another way with the help of digimer.<br>
&gt; I&#39;ve used the fence_virsh, from fence_agents.<br>
&gt;<br>
&gt; First of all I configured it in this way:<br>
&gt;<br>
</span>&gt; /primitive fence_server01 stonith:fence_virsh \<br>
&gt; /<br>
&gt; /        params ipaddr=virtnode01 port=server01.fqdn action=reboot<br>
&gt; login=root passwd=passwd delay=15  \/<br>
&gt; /        op monitor interval=60s /<br>
&gt; /primitive fence_server02 stonith:fence_virsh \/<br>
&gt; /        params ipaddr=virtnode02 port=server02.fqdn action=reboot<br>
&gt; login=root passwd=passwd delay=15  \/<br>
&gt; /        op monitor interval=60s /<br>
&gt; /<br>
&gt; /<br>
<span class="">&gt;<br>
&gt; But when I tried to fence a node, I received this errors:<br>
&gt;<br>
</span>&gt;  1.<br>
<span class="">&gt;     Jun 16 09:37:59 [1298] server01    pengine:  warning: pe_fence_node:<br>
&gt;         Node server02 will be fenced because p_fence_server01 is thought<br>
&gt;     to be active there<br>
</span>&gt;  2.<br>
<span class="">&gt;     Jun 16 09:37:59 [1299] server01       crmd:   notice: te_fence_node:<br>
&gt;         Executing reboot fencing operation (12) on server02 (timeout=60000)<br>
</span>&gt;  3.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:   notice:<br>
&gt;     handle_request:    Client crmd.1299.d339ea94 wants to fence (reboot)<br>
&gt;     &#39;server02&#39; with device &#39;(any)&#39;<br>
</span>&gt;  4.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:   notice:<br>
&gt;     initiate_remote_stonith_op:        Initiating remote operation<br>
&gt;     reboot for server02: 19fdb8e0-2611-45a7-b44d-b58fa0e99cab (0)<br>
</span>&gt;  5.<br>
<span class="">&gt;     Jun 16 09:37:59 [1297] server01      attrd:     info:<br>
&gt;     attrd_cib_callback:        Update 12 for probe_complete: OK (0)<br>
</span>&gt;  6.<br>
<span class="">&gt;     Jun 16 09:37:59 [1297] server01      attrd:     info:<br>
&gt;     attrd_cib_callback:        Update 12 for<br>
&gt;     probe_complete[server01]=true: OK (0)<br>
</span>&gt;  7.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:   notice:<br>
&gt;     can_fence_host_with_device:        p_fence_server02 can not fence<br>
&gt;     (reboot) server02: dynamic-list<br>
</span>&gt;  8.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:     info:<br>
&gt;     process_remote_stonith_query:      All queries have arrived,<br>
&gt;     continuing (1, 1, 1, 19fdb8e0-2611-45a7-b44d-b58fa0e99cab)<br>
</span>&gt;  9.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:   notice:<br>
&gt;     stonith_choose_peer:       Couldn&#39;t find anyone to fence server02<br>
&gt;     with &lt;any&gt;<br>
</span>&gt; 10.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:     info:<br>
&gt;     call_remote_stonith:       Total remote op timeout set to 60 for<br>
&gt;     fencing of node server02 for crmd.1299.19fdb8e0<br>
</span>&gt; 11.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:     info:<br>
&gt;     call_remote_stonith:       None of the 1 peers have devices capable<br>
&gt;     of terminating server02 for crmd.1299 (0)<br>
</span>&gt; 12.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:  warning:<br>
&gt;     get_xpath_object:  No match for //@st_delegate in /st-reply<br>
</span>&gt; 13.<br>
<span class="">&gt;     Jun 16 09:37:59 [1295] server01   stonithd:    error:<br>
&gt;     remote_op_done:    Operation reboot of server02 by server01 for<br>
&gt;     crmd.1299@server01.19fdb8e0: No such device<br>
</span>&gt; 14.<br>
<span class="">&gt;     Jun 16 09:37:59 [1299] server01       crmd:   notice:<br>
&gt;     tengine_stonith_callback:  Stonith operation<br>
&gt;     3/12:1:0:a989fb7b-1af1-4bac-992b-eef416e25775: No such device (-19)<br>
</span>&gt; 15.<br>
<span class="">&gt;     Jun 16 09:37:59 [1299] server01       crmd:   notice:<br>
&gt;     tengine_stonith_callback:  Stonith operation 3 for server02 failed<br>
&gt;     (No such device): aborting transition.<br>
</span>&gt; 16.<br>
<span class="">&gt;     Jun 16 09:37:59 [1299] server01       crmd:   notice:<br>
&gt;     abort_transition_graph:    Transition aborted: Stonith failed<br>
&gt;     (source=tengine_stonith_callback:697, 0)<br>
</span>&gt; 17.<br>
<span class="">&gt;     Jun 16 09:37:59 [1299] server01       crmd:   notice:<br>
&gt;     tengine_stonith_notify:    Peer server02 was not terminated (reboot)<br>
&gt;     by server01 for server01: No such device<br>
&gt;     (ref=19fdb8e0-2611-45a7-b44d-b58fa0e99cab) by client crmd.1299<br>
&gt;<br>
&gt;<br>
</span>&gt; So, I had to put *pcmk_host_list *parameter, like:<br>
<span class="">&gt;<br>
&gt; primitive fence_server01 stonith:fence_virsh \<br>
&gt;         params ipaddr=virtnode01 port=server01.fqdn action=reboot<br>
&gt; login=root passwd=passwd delay=15 pcmk_host_list=server01 \<br>
&gt;         op monitor interval=60s<br>
&gt; primitive fence_server02 stonith:fence_virsh \<br>
&gt;         params ipaddr=virtnode02 port=server02.fqdn action=reboot<br>
&gt; login=root passwd=passwd delay=15 pcmk_host_list=server02 \<br>
&gt;         op monitor interval=60s<br>
&gt;<br>
&gt; Could you explain me, why? I hope that this doesn&#39;t not sound rough,<br>
&gt; it&#39;s only I don&#39;t understand why.<br>
&gt;<br>
&gt; Thank you very much<br>
&gt; Oscar Salvador<br>
<br>
</span>Don&#39;t use &#39;delay=&quot;15&quot;&#39; on both nodes! It&#39;s means to give one node a<br>
head-start over the other to help avoid a &#39;dual fence&#39;. The node that<br>
has the delay will live while the node without a delay will die in a<br>
case where communications fails and both nodes try to fence the other at<br>
the same time.<br>
<br>
Say you have &#39;delay=&quot;15&quot;&#39; on &#39;server01&#39;; Both start to fence, server01<br>
looks up how to fence server02, sees no delay and immediately fences.<br>
Meanwhile, &#39;server02&#39; looks up how to fence &#39;server01&#39;, sees a delay and<br>
pauses. If server01 was really dead, after 15 seconds, it would proceed<br>
with the fence action. However, if server01 is alive, server02 will die<br>
long before it&#39;s pause expires.<br>
<span class="HOEnZb"><font color="#888888"><br></font></span></blockquote><div><br></div></div>Hey Digimer, I know, actually in my config I have only one &quot;delay&quot; specified for this purpose. Maybe was an copy/paste error.<br></div><div class="gmail_extra">Thanks anyway ;)<br><br></div><div class="gmail_extra">Oscar Salvador<br></div></div>