[Pacemaker] the instance id of clone is changed after removing constraints

[Andrew Beekhof]

2010/3/29 Junko IKEDA <ikedaj at intellilink.co.jp>:
> Hi,
>
> I have the following resources,
>
> ============
> Last updated: Mon Mar 29 16:09:53 2010
> Stack: openais
> Current DC: cspm01 - partition with quorum
> Version: 1.0.8-d353c41113af stable-1.0 tip
> 4 Nodes configured, 4 expected votes
> 2 Resources configured.
> ============
>
> Online: [ cspm01 cspm02 cspm03 cspm04 ]
>
>  Resource Group: group01
>     gDummy01   (ocf::pacemaker:Dummy): Started cspm01
>     gDummy02   (ocf::pacemaker:Dummy): Started cspm01
>     gDummy03   (ocf::pacemaker:Dummy): Started cspm01
>  Clone Set: clone01
>     Started: [ cspm01 cspm04 ]
>
> Operations:
> * Node cspm02:
> * Node cspm03:
> * Node cspm04:
>   cDummy01:1: migration-threshold=1
>    + (6) start: rc=0 (ok)
>    + (7) monitor: interval=10000ms rc=0 (ok)
> * Node cspm01:
>   gDummy01: migration-threshold=1
>    + (8) start: rc=0 (ok)
>    + (9) monitor: interval=10000ms rc=0 (ok)
>   gDummy02: migration-threshold=1
>    + (10) start: rc=0 (ok)
>    + (11) monitor: interval=10000ms rc=0 (ok)
>   gDummy03: migration-threshold=1
>    + (12) start: rc=0 (ok)
>    + (13) monitor: interval=10000ms rc=0 (ok)
>   cDummy01:0: migration-threshold=1
>    + (6) start: rc=0 (ok)
>    + (7) monitor: interval=10000ms rc=0 (ok)
>
>
> call "crm_resource" to move group01 from cspm01 to cspm04.
>
> # crm_resource -M -r group01 -H cspm04 -f -Q
>
> the above operation went well,
> but after trying to remove constraints depended on a move command,
>
> # crm_resource -U -r group01
>
> cDummy01:0 on cspm01 switched to cDummy01:1
> and
> cDummy01:1 on cspm04 switched to cDummy01:0
>
>
> # ptest -VVV -x pe-input-5.bz2
> ptest[21723]: 2010/03/29_16:50:56 WARN: unpack_nodes: Blind faith: not
> fencing unseen nodes
> ptest[21723]: 2010/03/29_16:50:56 notice: group_print:  Resource Group:
> group01
> ptest[21723]: 2010/03/29_16:50:56 notice: native_print:      gDummy01
> (ocf::pacemaker:Dummy): Started cspm04
> ptest[21723]: 2010/03/29_16:50:56 notice: native_print:      gDummy02
> (ocf::pacemaker:Dummy): Started cspm04
> ptest[21723]: 2010/03/29_16:50:56 notice: native_print:      gDummy03
> (ocf::pacemaker:Dummy): Started cspm04
> ptest[21723]: 2010/03/29_16:50:56 notice: clone_print:  Clone Set: clone01
> ptest[21723]: 2010/03/29_16:50:56 notice: short_print:      Started: [
> cspm01 cspm04 ]
> ptest[21723]: 2010/03/29_16:50:56 notice: RecurringOp:  Start recurring
> monitor (10s) for cDummy01:0 on cspm04
> ptest[21723]: 2010/03/29_16:50:56 notice: RecurringOp:  Start recurring
> monitor (10s) for cDummy01:1 on cspm01
> ptest[21723]: 2010/03/29_16:50:56 notice: LogActions: Leave resource
> gDummy01   (Started cspm04)
> ptest[21723]: 2010/03/29_16:50:56 notice: LogActions: Leave resource
> gDummy02   (Started cspm04)
> ptest[21723]: 2010/03/29_16:50:56 notice: LogActions: Leave resource
> gDummy03   (Started cspm04)
> ptest[21723]: 2010/03/29_16:50:56 notice: LogActions: Move resource
> cDummy01:0  (Started cspm01 -> cspm04)
> ptest[21723]: 2010/03/29_16:50:56 notice: LogActions: Move resource
> cDummy01:1  (Started cspm04 -> cspm01)
>
>
> globally_unique="false".
> Why the instance id of clone is changed?

Because it can.
Users of globally_unique clones do not get to care about the instance numbers.
They are for the cluster's internal use only.

Sorry.