PCS and CRM display the duplicates of nodes

[Alexey]

There are 3 controller nodes in pacemaker cluster. Here is output of "pcs status".
It shows "5 Nodes configured, 3 expected votes" and 3 online nodes, 2 offline nodes.

# pcs status
Cluster name:
Last updated: Thu Apr 16 15:56:31 2015
Last change: Thu Apr 16 15:55:33 2015 via crm_attribute on node-22.mng.kvh.ne.jp
Stack: classic openais (with plugin)
Current DC: node-17.mng.kvh.ne.jp - partition with quorum
Version: 1.1.10-14.el6_5.3-368c726
5 Nodes configured, 3 expected votes
32 Resources configured

Online: [ node-17.mng.kvh.ne.jp node-18.mng.kvh.ne.jp node-22.mng.kvh.ne.jp ]
OFFLINE: [ node-17 node-22 ]

Part of configuration.xml:
...........
<nodes>
  <node id="node-17.mng.kvh.ne.jp" uname="node-17.mng.kvh.ne.jp">
    <instance_attributes id="nodes-node-17.mng.kvh.ne.jp">
       <nvpair id="nodes-node-17.mng.kvh.ne.jp-gtid" name="gtid" value="d54fd46b4ee2-11e4-9ce1-7e4a74c78e0c:69850475"/>
    </instance_attributes>
  </node>
  <node id="node-22.mng.kvh.ne.jp" uname="node-22.mng.kvh.ne.jp">
    <instance_attributes id="nodes-node-22.mng.kvh.ne.jp">
      <nvpair id="nodes-node-22.mng.kvh.ne.jp-gtid" name="gtid" value="d54fd46b-4ee2-11e4-9ce1-7e4a74c78e0c:69850199"/>
    </instance_attributes>
  </node>
  <node id="node-18.mng.kvh.ne.jp" uname="node-18.mng.kvh.ne.jp">
    <instance_attributes id="nodes-node-18.mng.kvh.ne.jp">
      <nvpair id="nodes-node-18.mng.kvh.ne.jp-gtid" name="gtid" value="d54fd46b-4ee2-11e4-9ce1-7e4a74c78e0c:69850093"/>
    </instance_attributes>
  </node>
  <node id="node-22" uname="node-22"/>
  <node id="node-17" uname="node-17"/>
</nodes>
..........

[Alexey (aterekhin)]

Next case from another environment:

# crm status
Last updated: Fri Apr 17 06:41:43 2015
Last change: Thu Apr 16 14:45:16 2015 via cibadmin on node-38
Stack: openais
Current DC: node-41 - partition with quorum
Version: 1.1.8-f722cf1
6 Nodes configured, 3 expected votes
32 Resources configured.

Online: [ node-38 node-40 node-41 ]
OFFLINE: [ clone_p_neutron-plugin-openvswitch-agent p_neutron-dhcp-agent p_neutron-l3-agent ]

Part of "crm configure show":
............
node clone_p_neutron-plugin-openvswitch-agent
node node-38 \
        attributes standby="off"
node node-40 \
        attributes standby="off"
node node-41 \
        attributes standby="off"
node p_neutron-dhcp-agent
node p_neutron-l3-agent
...........

[Timur Nurlygayanov (tnurlygayanov)]

Hi,

what about

pcs cluster node remove node-17
pcs cluster node remove node-22

does it works or you have some issues with this command? Looks like you just have some incorrect nodes in your cloud and pacemaker/corosync can't resolve the hostnames.

[Timur Nurlygayanov (tnurlygayanov)]

By the way, if it is a bug we need to StepsTo Reproduce, because it is not clear now how to reproduce the issue on the test environment. All what we can see now - that it is MOS 5.x or MOS 6.0 releases (based on the version of corosync)

[Timur Nurlygayanov (tnurlygayanov)]

the status changed to Incomplete because it is not clear now how to reproduce the issue.
Alexey, colud you please provide more details about this case and steps how to reproduce the issue (because looks like you have two different environments with the same issue) and our dev team will fix this issue.

[Bogdan Dobrelya (bogdando)]

This is a known issue with adding nodes in corosync cluster