Prior to this update, the 'NeutronScale OCF' agent wasn't unable to provide specific clone status per host, due to a design flaw. Consequently, when starting only one clone in a host of the cluster, a status request for a different clone in the same host would also be positive; The startup status of all clones became unreliable, as that is the mechanism Pacemaker uses to start the clones.
This update addresses this issue by marking which specific clones are started in a host, and only returning an 'OK' status for the clones that are actually started.
As a result, all OpenStack Networking scale/neutron agent clones start reliably.
Description of problem:
The NeutronScale agent cannot accurately detect state due to inappropriate status file.
The current agent uses:
${HA_RSCTMP}/neutron-scale
However it needs to incorporate the ${OCF_RESOURCE_INSTANCE} variable somewhere.
Not doing so will make it appear to the cluster as if all instances are running there.
Version-Release number of selected component (if applicable):
openstack-neutron-2014.2.2-3.el7ost.noarch
How reproducible:
100%
Comment 3Fabio Massimo Di Nitto
2015-03-11 05:32:43 UTC
Created attachment 1000264[details]
make monitor action unique
Tested on rhos6-node1|2|3 cluster:
[root@rhos6-node1 neutron]# pcs status |grep -i scale
Clone Set: neutron-scale-clone [neutron-scale] (unique)
neutron-scale:0 (ocf::neutron:NeutronScale): Started rhos6-node1
neutron-scale:1 (ocf::neutron:NeutronScale): Started rhos6-node3
neutron-scale:2 (ocf::neutron:NeutronScale): Started rhos6-node2
[root@rhos6-node3 resource-agents]# pwd
/var/run/resource-agents
[root@rhos6-node3 resource-agents]# ls
Delay_ceilometer-delay neutron-scale-1
[root@rhos6-node1 resource-agents]# ls
Delay_ceilometer-delay neutron-scale-0
[root@rhos6-node2 resource-agents]# ls
Delay_ceilometer-delay neutron-scale-2
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
https://rhn.redhat.com/errata/RHBA-2015-0785.html
Description of problem: The NeutronScale agent cannot accurately detect state due to inappropriate status file. The current agent uses: ${HA_RSCTMP}/neutron-scale However it needs to incorporate the ${OCF_RESOURCE_INSTANCE} variable somewhere. Not doing so will make it appear to the cluster as if all instances are running there. Version-Release number of selected component (if applicable): openstack-neutron-2014.2.2-3.el7ost.noarch How reproducible: 100%