[Linux-HA] wti_nps stonith device monitor problem

Denis Chapligin chollya at satgate.net
Tue Jul 14 07:12:08 MDT 2009


Рш

On Mon, 13 Jul 2009 19:39:56 +0200
Dejan Muhamedagic <dejanmm at fastmail.fm> wrote:

> 
> Yes, it could be. stonithd are not coordinated, i.e. it may
> happen that two or more try to monitor the device at the same
> time. Another possibility is that the device is not very reliable
> and sometimes fails. I'd suggest to run just one stonith resource
> per device.
> 

So, with the only one copy of wti_nps stonith resource it continues to
stop:

Jul 14 13:37:44 lambda stonithd: [4824]: debug: Sending [/x,y#015] (len
5) 
Jul 14 13:37:44 lambda stonithd: [2406]: debug: Child process
wti_nps_Netreach4-PDS:1_monitor [4824] exited, its exit code: 0 when
signo=0. 
Jul 14 13:37:44 lambda stonithd: [2406]: debug:
Netreach4-PDS:1's (wti_nps) op monitor finished. op_result=0 
Jul 14
13:37:44 lambda stonithd: [2406]: debug: client STONITH_RA_EXEC_4823
(pid=4823) signed off 
Jul 14 13:38:47 lambda crmd: [2408]: debug:
cancel_op: Cancelling op 10 for Netreach4-PDS:1 (Netreach4-PDS:1:10)
Jul 14 13:38:47 lambda lrmd: [2405]: debug: cancel_op: operation
monitor[10] on stonith::wti_nps::Netreach4-PDS:1 for client 2408, its
parameters: CRM_meta_interval=[180000] CRM_meta_start_delay=[180000]
ipaddr=[85.195.135.252] CRM_meta_timeout=[360000] password=[fireball]
CRM_meta_clone_max=[2] crm_feature_set=[3.0.1]
CRM_meta_globally_unique=[false] CRM_meta_name=[monitor]
CRM_meta_clone=[1]  cancelled 
Jul 14 13:38:47 lambda crmd: [2408]:
info: do_lrm_rsc_op: Performing
key=14:11:0:7d20c040-cdd7-464d-9be5-5cc0fd1a348f
op=Netreach4-PDS:1_stop_0 ) 
Jul 14 13:38:47 lambda lrmd: [2405]: debug:
on_msg_perform_op: add an operation operation stop[15] on
stonith::wti_nps::Netreach4-PDS:1 for client 2408, its parameters:
CRM_meta_timeout=[20000] crm_feature_set=[3.0.1]  to the operation list


As you can see from the log above - monitor operation completed ok, but
crm decided to stop the resource.

-- 
						Denis Chapligin



More information about the Linux-HA mailing list