[Pacemaker] [Problem and Question] About negative setting of colocation.

[Andrew Beekhof]

On Wed, Jun 27, 2012 at 4:27 PM,  <renayama19661014 at ybb.ne.jp> wrote:
> Hi All,
>
> I registered these contents with Bugzilla.
>  * http://bugs.clusterlabs.org/show_bug.cgi?id=5074

Excellent, thankyou!
David and I have almost finished our "break everything and put it back
together" phase, we'll be diving back into bug reports in the coming
days.

>
> And I attached hb_report.
>
> Best Regards,
> Hideo Yamauchi.
>
> --- On Wed, 2012/6/27, renayama19661014 at ybb.ne.jp <renayama19661014 at ybb.ne.jp> wrote:
>
>> Hi All,
>>
>> We confirmed it about negative setting of colocation.
>>
>> It is colocation with with-rsc-role designation.
>>
>> We checked it in the next procedure.
>>
>> Step1) Start the first node. And we send cib.
>>
>> ============
>> Last updated: Wed Jun 27 22:58:15 2012
>> Stack: Heartbeat
>> Current DC: rh62-test1 (4a7c480c-ad95-4238-8506-7e5f75c551a4) - partition with quorum
>> Version: 1.0.12-unknown
>> 1 Nodes configured, unknown expected votes
>> 2 Resources configured.
>> ============
>>
>> Online: [ rh62-test1 ]
>>
>>  vipCheck       (ocf::pacemaker:Dummy): Started rh62-test1
>>  Master/Slave Set: msPostgresql
>>      Masters: [ rh62-test1 ]
>>      Stopped: [ postgresql:1 ]
>>
>> Node Attributes:
>> * Node rh62-test1:
>>     + master-postgresql:0               : 10
>>
>> Migration summary:
>> * Node rh62-test1:
>>
>>
>> Step2) Start the second node next.
>>
>> [root at rh62-test1 ~]# ptest -x pe-input-5 -VVV
>> ptest[29213]: 2012/06/27_23:30:43 notice: unpack_config: On loss of CCM Quorum: Ignore
>> ptest[29213]: 2012/06/27_23:30:43 WARN: unpack_nodes: Blind faith: not fencing unseen nodes
>> ptest[29213]: 2012/06/27_23:30:43 notice: native_print: vipCheck        (ocf::pacemaker:Dummy): Started rh62-test1
>> ptest[29213]: 2012/06/27_23:30:43 notice: clone_print:  Master/Slave Set: msPostgresql
>> ptest[29213]: 2012/06/27_23:30:43 notice: short_print:      Masters: [ rh62-test1 ]
>> ptest[29213]: 2012/06/27_23:30:43 notice: short_print:      Stopped: [ postgresql:1 ]
>> ptest[29213]: 2012/06/27_23:30:43 notice: RecurringOp:  Start recurring monitor (10s) for postgresql:0 on rh62-test2
>> ptest[29213]: 2012/06/27_23:30:43 notice: RecurringOp:  Start recurring monitor (9s) for postgresql:1 on rh62-test1
>> ptest[29213]: 2012/06/27_23:30:43 notice: RecurringOp:  Start recurring monitor (10s) for postgresql:0 on rh62-test2
>> ptest[29213]: 2012/06/27_23:30:43 notice: RecurringOp:  Start recurring monitor (9s) for postgresql:1 on rh62-test1
>> ptest[29213]: 2012/06/27_23:30:43 notice: LogActions: Leave   resource vipCheck (Started rh62-test1)
>> ptest[29213]: 2012/06/27_23:30:43 notice: LogActions: Demote  postgresql:0      (Master -> Slave rh62-test1)
>> ptest[29213]: 2012/06/27_23:30:43 notice: LogActions: Stop    resource postgresql:0     (rh62-test1)
>> ptest[29213]: 2012/06/27_23:30:43 notice: LogActions: Start   postgresql:0      (rh62-test2)
>> ptest[29213]: 2012/06/27_23:30:43 notice: LogActions: Start   postgresql:1      (rh62-test1)
>> ptest[29213]: 2012/06/27_23:30:43 notice: LogActions: Promote postgresql:1      (Stopped -> Master rh62-test1)
>>
>>
>> However, as for the "postgresql:0" resource, it is done Demote.
>> And replaced with "postgresql:1" resource and am done Promote.
>> This replacement has a problem by useless movement.
>>
>> And the cluster repeats Promote,Demote after this and does not constitute a resource definitely.
>>
>> Does it work definitely to set -INFINITY by with-rsc-role designation in colocation?
>> Is it a mistake to appoint it in specifications?
>>
>>       <rsc_colocation id="rsc_colocation-7" rsc="msPostgresql" rsc-role="Master" score="INFINITY" with-rsc="vipCheck"/>
>>       <rsc_colocation id="rsc_colocation-8" rsc="vipCheck" score="-INFINITY" with-rsc="msPostgresql" with-rsc-role="Slave"/>
>>       <rsc_order first="vipCheck" first-action="start" id="rsc_order-7" score="INFINITY" then="msPostgresql" then-action="promote"/>
>>
>> Or is this a bug?
>>
>>  * I confirmed it in ptest of Pacemaker1.1.7, but the result was the same.
>>
>> Best Regards,
>> Hideo Yamauchi.
>
> _______________________________________________
> 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