[Pacemaker] [Problem] Though order limitation exists, start which is not carried out is made.

Andrew Beekhof andrew at beekhof.net
Wed Aug 15 10:06:00 UTC 2012


On Fri, Jul 6, 2012 at 11:38 AM,  <renayama19661014 at ybb.ne.jp> wrote:
> Hi All,
>
> We performed setting avoiding the next problem.
>
>  * http://www.gossamer-threads.com/lists/linuxha/pacemaker/80250
>   -> http://bugs.clusterlabs.org/show_bug.cgi?id=5070
>
>  * http://www.gossamer-threads.com/lists/linuxha/pacemaker/80549
>   -> http://bugs.clusterlabs.org/show_bug.cgi?id=5075
>
> The resource finally started in Master node definitely.
>
> ----------------------------------------------------------------------------------------------------------------
> ============
> Last updated: Fri Jul  6 10:19:42 2012
> Stack: Heartbeat
> Current DC: sr1 (31411278-b503-4e17-9cf1-38638babb4ff) - partition with quorum
> Version: 1.0.12-unknown
> 1 Nodes configured, unknown expected votes
> 7 Resources configured.
> ============
>
> Online: [ sr1 ]
>
>  vipCheck       (ocf::heartbeat:VIPcheck):      Started sr1
>  vipCheck2      (ocf::pacemaker:Dummy): Started sr1
>  vip-slave      (ocf::heartbeat:IPaddr2):       Started sr1
>  Resource Group: master-group
>      vip-master (ocf::heartbeat:IPaddr2):       Started sr1
>      vip-rep    (ocf::heartbeat:IPaddr2):       Started sr1
>  Master/Slave Set: msPostgresql
>      Masters: [ sr1 ]
>      Stopped: [ postgresql:1 ]
>  Clone Set: clnPingd
>      Started: [ sr1 ]
>  Clone Set: clnDiskd1
>      Started: [ sr1 ]
> ----------------------------------------------------------------------------------------------------------------
>
>
> But start of vipCheck2 is made idly when we watch progress on the way.
>  * Start of this vipCheck2 is really state transition of the red limit that is not carried out.
>  * It exists in "pe-input-1.bz2".
>
> ----------------------------------------------------------------------------------------------------------------
> (snip)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: RecurringOp:  Start recurring monitor (10s) for prmPingd:0 on sr1
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: RecurringOp:  Start recurring monitor (10s) for prmDiskd1:0 on sr1
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Leave   resource vipCheck      (Stopped)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Start   vipCheck2      (sr1)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Leave   resource vip-slave     (Stopped)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Leave   resource vip-master    (Stopped)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Leave   resource vip-rep       (Stopped)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Leave   resource postgresql:0  (Stopped)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Leave   resource postgresql:1  (Stopped)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Start   prmPingd:0     (sr1)
> Jul  6 09:52:52 sr1 pengine: [2771]: notice: LogActions: Start   prmDiskd1:0    (sr1)
> Jul  6 09:52:52 sr1 crmd: [2766]: info: do_state_transition: State transition S_POLICY_ENGINE -> S_TRANSITION_ENGINE [ input=I_PE_SUCCESS cause=C_IPC_MESSAGE origin=handle_response ]
> Jul  6 09:52:52 sr1 crmd: [2766]: info: unpack_graph: Unpacked transition 1: 20 actions in 20 synapses
> Jul  6 09:52:52 sr1 crmd: [2766]: info: do_te_invoke: Processing graph 1 (ref=pe_calc-dc-1341535972-14) derived from /var/lib/pengine/pe-input-1.bz2
> (snip)
> ----------------------------------------------------------------------------------------------------------------
>
>
> Because there is limitation of vipCheck and vipCheck2, we think that the state transition of start of this useless vipCheck2 should not be made.
>
> (snip)
>       <rsc_order first="vipCheck" id="rsc_order-7" score="INFINITY" symmetrical="false" then="vipCheck2"/>
> (snip)
>
>
> Is this specifications?
> Or is it a bug?
>
>  * I registered a problem with Bugzilla.
>   * http://bugs.clusterlabs.org/show_bug.cgi?id=5079

Excellent.  I'll follow up there (likewise for the other bugs you've
sent to the list recently :-)




More information about the Pacemaker mailing list