[Pacemaker] RA Supporting the reload action

Felix Zachlod fz.lists at sis-gmbh.info
Thu Apr 11 13:47:50 UTC 2013


Hello gain folks,

I have been implementing a reload action for a resource agent. The action is
advertised in the meta-data and it works when invoking it directly.
According to http://linux-ha.org/wiki/OCF_Resource_Agents and
http://clusterlabs.org/doc/en-US/Pacemaker/1.0/html/Pacemaker_Explained/s-re
load.html I assumed that the crm should trigger reload after a modification
of the resource params.

"reload - reload the configuration (non-unique parameters only) of the
resource instance without disrupting the service"

But after modifying a non-unique param the pacemaker decided to restart my
resource. 

Apr 11 15:31:01 storage-test-a crmd: [7177]: info: te_rsc_command:
Initiating action 58: stop p_lun_vmfs_test_stop_0 on storage-test-b
Apr 11 15:31:01 storage-test-a attrd: [7176]: notice: attrd_trigger_update:
Sending flush op to all hosts for: last-failure-p_drbd0:0 (1365582777)
Apr 11 15:31:01 storage-test-a pengine: [7563]: notice: process_pe_message:
Transition 300: PEngine Input stored in: /var/lib/pengine/pe-input-285.bz2
Apr 11 15:31:01 storage-test-a attrd: [7176]: notice: attrd_trigger_update:
Sending flush op to all hosts for: probe_complete (true)
Apr 11 15:31:02 storage-test-a crmd: [7177]: info: te_rsc_command:
Initiating action 2: start p_lun_vmfs_test_start_0 on storage-test-b

Is there anything I am missing about the reload thing?

Will the crm use reload even in case there are order constraints for other
resources that depend on this RA?

Thank you all

Regards, Felix





More information about the Pacemaker mailing list