[Pacemaker] Question regarding starting of master/slave resources and ELECTIONs

Thu Apr 14 08:49:09 UTC 2011

On Wed, Apr 13, 2011 at 7:19 PM, Bob Schatz <bschatz at yahoo.com> wrote:
> Andrew,
> Thanks for responding.  Comments inline with <Bob>
> ________________________________
> From: Andrew Beekhof <andrew at beekhof.net>
> To: The Pacemaker cluster resource manager <pacemaker at oss.clusterlabs.org>
> Cc: Bob Schatz <bschatz at yahoo.com>
> Sent: Tue, April 12, 2011 11:23:14 PM
> Subject: Re: [Pacemaker] Question regarding starting of master/slave
> resources and ELECTIONs
>
> On Wed, Apr 13, 2011 at 4:54 AM, Bob Schatz <bschatz at yahoo.com> wrote:
>> Hi,
>> I am running Pacemaker 1.0.9 with Heartbeat 3.0.3.
>> I create 5 master/slave resources in /etc/ha.d/resource.d/startstop during
>> post-start.
>
> I had no idea this was possible.  Why would you do this?
> <Bob>  We and I know of a couple of other companies, bundle
> LinuxHA/Pacemaker into an appliance.  For me, when the appliance boots, it
> creates HA resources based on the hardware it discovers.   I assumed that
> once POST-START was called in the startstop script and we have a DC then the
> cluster is up and running.  I then use "crm" commands to create the
> configuration, etc.  I further assumed that since we have one DC in the
> cluster then all "crm" commands which modify the configuration would be
> ordered even if the DC fails over to a different node.  Is this incorrect?

Its correct. Its just a usecase I had ever thought of :-)

>
>> I noticed that 4 of the master/slave resources will start right away but
>> the
>> 5 master/slave resource seems to take a minute or so and I am only running
>> with one node.
>> Is this expected?
>
> Probably, if the other 4 take around a minute each to start.
> There is an lrmd config variable that controls how much parallelism it
> allows (but i forget the name).
> <Bob> It's max-children and I set it to 40 for this test to see if it would
> change the behavior.  (/sbin/lrmadmin -p max-children 40)

Thats surprising.  I'll have a look at the logs.

>
>> My configuration is below and I have also attached ha-debug.
>> Also, what triggers a crmd election?
>
> Node up/down events and whenever someone replaces the cib (which the
> shell used to do a lot).
> <Bob> For my test, I only started one node so that I could avoid node
> up/down events.  I believe the log shows the cib being replaced.  Since I am
> using crm then I assume it must be due to crm.   Do the crm_resource, etc
> commands also replace the cib?

No.  Also I believe recent versions of the shell no longer use a
replace operation by default.

> Would that avoid elections as a result of
> cibs being replaced?

Only avoiding the replace operation in the first place.
After a replace we have to repopulate the status section to ensure it
is accurate.

> Thanks,
> Bob
>
>>  I seemed to have a lot of elections in
>> the attached log.  I was assuming that on a single node I would only run
>> the
>> election once in the beginning and then there would not be another one
>> until
>> a new node joined.
>>
>> Thanks,
>> Bob
>>
>> My configuration is:
>> node $id="856c1f72-7cd1-4906-8183-8be87eef96f2" mgraid-s000030311-1
>> primitive SSJ000030312 ocf:omneon:ss \
>>         params ss_resource="SSJ000030312"
>> ssconf="/var/omneon/config/config.J000030312" \
>>         op monitor interval="3s" role="Master" timeout="7s" \
>>         op monitor interval="10s" role="Slave" timeout="7" \
>>         op stop interval="0" timeout="20" \
>>         op start interval="0" timeout="300"
>> primitive SSJ000030313 ocf:omneon:ss \
>>         params ss_resource="SSJ000030313"
>> ssconf="/var/omneon/config/config.J000030313" \
>>         op monitor interval="3s" role="Master" timeout="7s" \
>>         op monitor interval="10s" role="Slave" timeout="7" \
>>         op stop interval="0" timeout="20" \
>>         op start interval="0" timeout="300"
>> primitive SSJ000030314 ocf:omneon:ss \
>>         params ss_resource="SSJ000030314"
>> ssconf="/var/omneon/config/config.J000030314" \
>>         op monitor interval="3s" role="Master" timeout="7s" \
>>         op monitor interval="10s" role="Slave" timeout="7" \
>>         op stop interval="0" timeout="20" \
>>         op start interval="0" timeout="300"
>> primitive SSJ000030315 ocf:omneon:ss \
>>         params ss_resource="SSJ000030315"
>> ssconf="/var/omneon/config/config.J000030315" \
>>         op monitor interval="3s" role="Master" timeout="7s" \
>>         op monitor interval="10s" role="Slave" timeout="7" \
>>         op stop interval="0" timeout="20" \
>>         op start interval="0" timeout="300"
>> primitive SSS000030311 ocf:omneon:ss \
>>         params ss_resource="SSS000030311"
>> ssconf="/var/omneon/config/config.S000030311" \
>>         op monitor interval="3s" role="Master" timeout="7s" \
>>         op monitor interval="10s" role="Slave" timeout="7" \
>>         op stop interval="0" timeout="20" \
>>         op start interval="0" timeout="300"
>> primitive icms lsb:S53icms \
>>         op monitor interval="5s" timeout="7" \
>>         op start interval="0" timeout="5"
>> primitive mgraid-stonith stonith:external/mgpstonith \
>>         params hostlist="mgraid-canister" \
>>         op monitor interval="0" timeout="20s"
>> primitive omserver lsb:S49omserver \
>>         op monitor interval="5s" timeout="7" \
>>         op start interval="0" timeout="5"
>> ms ms-SSJ000030312 SSJ000030312 \
>>         meta clone-max="2" notify="true" globally-unique="false"
>> target-role="Started"
>> ms ms-SSJ000030313 SSJ000030313 \
>>         meta clone-max="2" notify="true" globally-unique="false"
>> target-role="Started"
>> ms ms-SSJ000030314 SSJ000030314 \
>>         meta clone-max="2" notify="true" globally-unique="false"
>> target-role="Started"
>> ms ms-SSJ000030315 SSJ000030315 \
>>         meta clone-max="2" notify="true" globally-unique="false"
>> target-role="Started"
>> ms ms-SSS000030311 SSS000030311 \
>>         meta clone-max="2" notify="true" globally-unique="false"
>> target-role="Started"
>> clone Fencing mgraid-stonith
>> clone cloneIcms icms
>> clone cloneOmserver omserver
>> location ms-SSJ000030312-master-w1 ms-SSJ000030312 \
>>         rule $id="ms-SSJ000030312-master-w1-rule" $role="master" 100:
>> #uname
>> eq mgraid-s000030311-0
>> location ms-SSJ000030313-master-w1 ms-SSJ000030313 \
>>         rule $id="ms-SSJ000030313-master-w1-rule" $role="master" 100:
>> #uname
>> eq mgraid-s000030311-0
>> location ms-SSJ000030314-master-w1 ms-SSJ000030314 \
>>         rule $id="ms-SSJ000030314-master-w1-rule" $role="master" 100:
>> #uname
>> eq mgraid-s000030311-0
>> location ms-SSJ000030315-master-w1 ms-SSJ000030315 \
>>         rule $id="ms-SSJ000030315-master-w1-rule" $role="master" 100:
>> #uname
>> eq mgraid-s000030311-0
>> location ms-SSS000030311-master-w1 ms-SSS000030311 \
>>         rule $id="ms-SSS000030311-master-w1-rule" $role="master" 100:
>> #uname
>> eq mgraid-s000030311-0
>> order orderms-SSJ000030312 0: cloneIcms ms-SSJ000030312
>> order orderms-SSJ000030313 0: cloneIcms ms-SSJ000030313
>> order orderms-SSJ000030314 0: cloneIcms ms-SSJ000030314
>> order orderms-SSJ000030315 0: cloneIcms ms-SSJ000030315
>> order orderms-SSS000030311 0: cloneIcms ms-SSS000030311
>> property $id="cib-bootstrap-options" \
>>         dc-version="1.0.9-89bd754939df5150de7cd76835f98fe90851b677" \
>>         cluster-infrastructure="Heartbeat" \
>>         dc-deadtime="5s" \
>>         stonith-enabled="true"
>>
>>
>> _______________________________________________
>> 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://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>>
>>
>
>