Hi guys.
I have clone pgsql resource which - if its notify interval
!= 0s - floods pacemaker logs with:
...
error: clone_op_key: Triggered fatal assertion at
pcmk_graph_producer.c:207 : (n_type != NULL) && (n_task != NULL)
error: pcmk__notify_key: Triggered fatal assertion at
actions.c:340 : op_type != NULL
error: clone_op_key: Triggered fatal assertion at
pcmk_graph_producer.c:207 : (n_type != NULL) && (n_task != NULL)
error: pcmk__notify_key: Triggered fatal assertion at
actions.c:340 : op_type != NULL
...
error: pcmk__create_history_xml: Triggered fatal assertion
at pcmk_sched_actions.c:1164 : n_type != NULL
error: pcmk__create_history_xml: Triggered fatal assertion
at pcmk_sched_actions.c:1165 : n_task != NULL
error: pcmk__notify_key: Triggered fatal assertion at
actions.c:340 : op_type != NULL
...
In other words resource must be:
-> $ pcs resource update PGSQL op notify interval=0s
in order to errors disappear.
Would you have any thoughts as to what is (not)happening there?
(btw. in case it might matter - I do not use, do not have
set any 'alerts')
If you use a resource agent which uses/allows 'notify' in
'operations' - can you use different 'interval' values?
Without getting those errors, naturally.
Also, cluster - except for the errors in the log - seems
healthy & operates a ok.
many thanks, L.
_______________________________________________
Manage your subscription:
https://lists.clusterlabs.org/mailman/listinfo/users
ClusterLabs home: https://www.clusterlabs.org/
