[Pacemaker] Resources move on Pacemaker + Corosync cluster with set stickiness

Andrew Beekhof andrew at beekhof.net
Tue May 27 21:13:05 UTC 2014


On 27 May 2014, at 8:23 pm, Danilo Malcangio <d.malcangio at eletech.it> wrote:

> I've removed the location constraint and it seems the resources don't move anymore if I reboot BX-1. 
> During reboot I noticed on crm_mon that resources for one second appeared offline and then they stayed on BX-2. Does anyone know why that happened?

This could just be an artefact of the status section being refreshed - which happens as a delete + repopulate.
On the other hand, it could also be us recovering from a situation where the resource was active on both nodes.

grep for pengine in the logs - that will tell you if anything was moved/restarted.

> 
> I've tried reconfiguring my cluster following this official guide (http://clusterlabs.org/doc/en-US/Pacemaker/1.1-plugin/html/Clusters_from_Scratch/index.html) 5. Creating an Active/Passive Cluster
> but I still get the same problem, at reboot nodes go back to the preferred node BX-1 that is exactly the contrary of what is stated in the guide.
> 
> Still wondering how the location constraint and stickiness work.
> 
>> Try crm_mon -o and look for monitor operations that return 0 instead of 7
>> 
> 
> I am sorry Andrew, I've tried to follow your advice but didn't quite catch what you want me to look for :(

Failing monitor actions that indicate the resource was already started without pacemaker telling it to do so.

> 
> 
> Il 23/05/2014 2.42, Andrew Beekhof ha scritto:
>> On 22 May 2014, at 9:00 pm, Danilo Malcangio <d.malcangio at eletech.it>
>>  wrote:
>> 
>> 
>>> Hi Andrew, first of all thanks for answering.
>>> 
>>> 
>>>> Almost certainly the node is configured to start those resources at bootup.
>>>> Don't do that :)
>>>> 
>>>> 
>>> Are you advicing me to delete the location constraint? (location prefer-et-ipbx-1 cluster-group 100: BX-1) 
>>> Or is it something else that starts the resources on BX-1 Node?? 
>>> 
>> By the looks of it, yes.
>> Try crm_mon -o and look for monitor operations that return 0 instead of 7
>> 
>> 
>>> I've removed all the scripts from the startup sequence with "update-rc.d -f  SCRIPT_NAME remove"
>>> 
>>> Thank you
>>> 
>>> Il 22/05/2014 12.25, Andrew Beekhof ha scritto:
>>> 
>>>> On 22 May 2014, at 5:31 pm, Danilo Malcangio <d.malcangio at eletech.it>
>>>> 
>>>>  wrote:
>>>> 
>>>> 
>>>> 
>>>>> Hi everyone,
>>>>> I've created an active/passive 2 node cluster following the documentation on clusterlabs.
>>>>> My cluster has the following characteristics
>>>>> Debian Wheezy 7.2.0
>>>>> Pacemaker 1.1.7
>>>>> Corosync 1.4.2
>>>>> 
>>>>> I've made it with the following configuration
>>>>> 
>>>>> node BX-1
>>>>> node BX-2
>>>>> primitive cluster-apache2 ocf:heartbeat:apache \
>>>>>     params configfile="/etc/apache2/apache2.conf" httpd="/usr/sbin/apache2" port="80" \
>>>>>     op monitor interval="10s" timeout="60s" \
>>>>>     op start interval="0" timeout="40s" \
>>>>>     op stop interval="0" timeout="60s"
>>>>> primitive cluster-asterisk lsb:asterisk \
>>>>>     op monitor interval="30" \
>>>>>     op start interval="0" timeout="120s" \
>>>>>     op stop interval="0" timeout="120s"
>>>>> primitive cluster-ip ocf:heartbeat:IPaddr2 \
>>>>>     params ip="10.2.30.10" cidr_netmask="20" \
>>>>>     op monitor interval="10s" timeout="20s" \
>>>>>     op start interval="0" timeout="20s" \
>>>>>     op stop interval="0" timeout="20s"
>>>>> primitive cluster-ntp lsb:ntp \
>>>>>     op monitor interval="30" \
>>>>>     op start interval="0" timeout="120s" \
>>>>>     op stop interval="0" timeout="120s"
>>>>> primitive cluster-tftp lsb:tftpd-hpa \
>>>>>     op monitor interval="30" \
>>>>>     op start interval="0" timeout="120s" \
>>>>>     op stop interval="0" timeout="120s"
>>>>> group cluster-group cluster-ip cluster-asterisk cluster-apache2 cluster-tftp cluster-ntp \
>>>>>     meta resource-stickiness="101"
>>>>> location prefer-et-ipbx-1 cluster-group 100: BX-1
>>>>> colocation cluster-dependency inf: cluster-ip cluster-asterisk cluster-apache2 cluster-tftp cluster-ntp
>>>>> property $id="cib-bootstrap-options" \
>>>>>     dc-version="1.1.7-ee0730e13d124c3d58f00016c3376a1de5323cff" \
>>>>>     cluster-infrastructure="openais" \
>>>>>     expected-quorum-votes="2" \
>>>>>     stonith-enabled="false" \
>>>>>     no-quorum-policy="ignore" \
>>>>>     default-resource-stickiness="1"
>>>>> rsc_defaults $id="rsc-options" \
>>>>>     failure-timeout="60s"
>>>>> 
>>>>> I've set location constraint to have BX-1 as the preferred node at cluster startup, and group stickiness at 101 to avoid moving resources when the master node comes back up (following this guide 
>>>>> 
>>>>> http://foaa.de/old-blog/2010/10/intro-to-pacemaker-part-2-advanced-topics/trackback/index.html
>>>>> 
>>>>> ).
>>>>> 
>>>>> I've got the following problem: resources move when I reboot nodes.
>>>>> 
>>>>> If i stop corosync on BX-1 resources move to BX-2 and when i restart corosync on BX-1 they stay on BX-2 (as I expected). 
>>>>> But all this doesn't happen with reboot (when I reboot BX-1).
>>>>> 
>>>>> 
>>>> Almost certainly the node is configured to start those resources at bootup.
>>>> Don't do that :)
>>>> 
>>>> 
>>>> 
>>>>> After the reboot of BX-1 resources move to BX-2 and when BX-1 comes back up the resources move back to BX-1.
>>>>> 
>>>>> What am I missing in the configuration??
>>>>> 
>>>>> Thank you very much for the support
>>>>> 
>>>>> 
>>>>> 
>>>>> P.s. I attach also the allocation scores obtained with ptest -sL
>>>>> 
>>>>> BX-1 has resources
>>>>> 
>>>>> Allocation scores:
>>>>> group_color: et-cluster allocation score on ET-IPBX-1: 100
>>>>> group_color: et-cluster allocation score on ET-IPBX-2: 0
>>>>> group_color: cluster-ip allocation score on ET-IPBX-1: 201
>>>>> group_color: cluster-ip allocation score on ET-IPBX-2: 0
>>>>> group_color: cluster-asterisk allocation score on ET-IPBX-1: 101
>>>>> group_color: cluster-asterisk allocation score on ET-IPBX-2: 0
>>>>> group_color: cluster-apache2 allocation score on ET-IPBX-1: 101
>>>>> group_color: cluster-apache2 allocation score on ET-IPBX-2: 0
>>>>> group_color: cluster-tftp allocation score on ET-IPBX-1: 101
>>>>> group_color: cluster-tftp allocation score on ET-IPBX-2: 0
>>>>> group_color: cluster-ntp allocation score on ET-IPBX-1: 101
>>>>> group_color: cluster-ntp allocation score on ET-IPBX-2: 0
>>>>> native_color: cluster-ip allocation score on ET-IPBX-1: 3231
>>>>> native_color: cluster-ip allocation score on ET-IPBX-2: 0
>>>>> native_color: cluster-asterisk allocation score on ET-IPBX-1: 1515
>>>>> native_color: cluster-asterisk allocation score on ET-IPBX-2: -INFINITY
>>>>> native_color: cluster-apache2 allocation score on ET-IPBX-1: 707
>>>>> native_color: cluster-apache2 allocation score on ET-IPBX-2: -INFINITY
>>>>> native_color: cluster-tftp allocation score on ET-IPBX-1: 303
>>>>> native_color: cluster-tftp allocation score on ET-IPBX-2: -INFINITY
>>>>> native_color: cluster-ntp allocation score on ET-IPBX-1: 101
>>>>> native_color: cluster-ntp allocation score on ET-IPBX-2: -INFINITY
>>>>> 
>>>>> BX-2 has resources (while BX-1 reboot)
>>>>> 
>>>>> Allocation scores:
>>>>> group_color: et-cluster allocation score on ET-IPBX-1: -INFINITY
>>>>> group_color: et-cluster allocation score on ET-IPBX-2: 0
>>>>> group_color: cluster-ip allocation score on ET-IPBX-1: -INFINITY
>>>>> group_color: cluster-ip allocation score on ET-IPBX-2: 101
>>>>> group_color: cluster-asterisk allocation score on ET-IPBX-1: 0
>>>>> group_color: cluster-asterisk allocation score on ET-IPBX-2: 101
>>>>> group_color: cluster-apache2 allocation score on ET-IPBX-1: 0
>>>>> group_color: cluster-apache2 allocation score on ET-IPBX-2: 101
>>>>> group_color: cluster-tftp allocation score on ET-IPBX-1: 0
>>>>> group_color: cluster-tftp allocation score on ET-IPBX-2: 101
>>>>> group_color: cluster-ntp allocation score on ET-IPBX-1: 0
>>>>> group_color: cluster-ntp allocation score on ET-IPBX-2: 101
>>>>> native_color: cluster-ip allocation score on ET-IPBX-1: -INFINITY
>>>>> native_color: cluster-ip allocation score on ET-IPBX-2: 3131
>>>>> native_color: cluster-asterisk allocation score on ET-IPBX-1: -INFINITY
>>>>> native_color: cluster-asterisk allocation score on ET-IPBX-2: 1515
>>>>> native_color: cluster-apache2 allocation score on ET-IPBX-1: -INFINITY
>>>>> native_color: cluster-apache2 allocation score on ET-IPBX-2: 707
>>>>> native_color: cluster-tftp allocation score on ET-IPBX-1: -INFINITY
>>>>> native_color: cluster-tftp allocation score on ET-IPBX-2: 303
>>>>> native_color: cluster-ntp allocation score on ET-IPBX-1: -INFINITY
>>>>> native_color: cluster-ntp allocation score on ET-IPBX-2: 101
>>>>> _______________________________________________
>>>>> Pacemaker mailing list: 
>>>>> 
>>>>> Pacemaker at oss.clusterlabs.org
>>>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>>>> 
>>>>> 
>>>>> 
>>>>> Project Home: 
>>>>> 
>>>>> http://www.clusterlabs.org
>>>>> 
>>>>> 
>>>>> Getting started: 
>>>>> 
>>>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>>>> 
>>>>> 
>>>>> Bugs: 
>>>>> 
>>>>> http://bugs.clusterlabs.org
>>>> 
>>>> _______________________________________________
>>>> Pacemaker mailing list: 
>>>> 
>>>> Pacemaker at oss.clusterlabs.org
>>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>>> 
>>>> 
>>>> 
>>>> Project Home: 
>>>> 
>>>> http://www.clusterlabs.org
>>>> 
>>>> 
>>>> Getting started: 
>>>> 
>>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>>> 
>>>> 
>>>> Bugs: 
>>>> 
>>>> http://bugs.clusterlabs.org
>>> _______________________________________________
>>> Pacemaker mailing list: 
>>> Pacemaker at oss.clusterlabs.org
>>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>> 
>>> 
>>> Project Home: 
>>> http://www.clusterlabs.org
>>> 
>>> Getting started: 
>>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>>> 
>>> Bugs: 
>>> http://bugs.clusterlabs.org
>> 
>> 
>> _______________________________________________
>> Pacemaker mailing list: 
>> Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>> 
>> 
>> Project Home: 
>> http://www.clusterlabs.org
>> 
>> Getting started: 
>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> 
>> Bugs: 
>> http://bugs.clusterlabs.org
> 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20140528/46e83065/attachment-0004.sig>


More information about the Pacemaker mailing list