Andrew Beekhof <[EMAIL PROTECTED]> writes: > On Dec 7, 2007, at 11:56 AM, Keisuke MORI wrote: > >> Andrew, >> >> Can I ask a question about the internal status of the PE? >> >> My SNMP subagent code is using cluster_status(pe_working_set_t) >> to analyze the current status of resources like crm_mon. >> >> When a parent resource(group/clone/master) is unmanaged, >> 'running_on' and 'allowed_nodes' member of resource_t gets NULL. >> Is this an expected value? > > I thought that group/clone/master always had NULL... since they can be > running on more than one node (especially clone and m/s resources)
Judging from the output of the SNMP agent, two pair of a clone and a primitive are observed and each of the parent clone has the node which its child primitive is running on. Maybe my code is doing something wrong, I'll check it again. > > I recall also doing something special for unmanaged resources but I > can probably change that behavior for you. > > that said, it would be better to use the recently added API call: > node_t *(*location)(resource_t *, GListPtr*, gboolean); > > eg. > node_t *native_location(resource_t *rsc, GListPtr *list, gboolean > current) > Thanks, I will look into this. -- Keisuke MORI NTT DATA Intellilink Corporation _______________________________________________________ Linux-HA-Dev: Linux-HA-Dev@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev Home Page: http://linux-ha.org/
