High Availability クラスターでノードがパニック状態になると、fence_aws のフェンスアクションが "Timed out waiting to power OFF" および "Unable to obtain correct plug status or plug is not available" で失敗する

Issue

• AWS 上の Pacemaker クラスターノードでカーネルパニックが発生したとき、または echo c > /proc/sysrq-trigger を実行してクラッシュさせたとき、そのノードに対するフェンスアクションが "Timed out waiting to power OFF" で失敗します。アクションを再試行すると、"Unable to obtain correct plug status or plug is not available" で繰り返し失敗します。
• pcs stonith fence コマンドを使用してノードを手動でフェンスすることは成功します。
• この問題は断続的に発生することがあります。

Apr  2 22:18:04 ip-10-0-0-17 corosync[14259]: [TOTEM ] A processor failed, forming new configuration.
Apr  2 22:18:05 ip-10-0-0-17 corosync[14259]: [TOTEM ] A new membership (10.0.0.17:177) was formed. Members left: 2
Apr  2 22:18:05 ip-10-0-0-17 corosync[14259]: [TOTEM ] Failed to receive the leave message. failed: 2
Apr  2 22:18:05 ip-10-0-0-17 corosync[14259]: [CPG   ] downlist left_list: 1 received
Apr  2 22:18:05 ip-10-0-0-17 corosync[14259]: [QUORUM] Members[1]: 1
Apr  2 22:18:05 ip-10-0-0-17 corosync[14259]: [MAIN  ] Completed service synchronization, ready to provide service.
Apr  2 22:18:05 ip-10-0-0-17 pacemakerd[14283]:  notice: Node node2 state is now lost
...
Apr  2 22:18:06 ip-10-0-0-17 crmd[14289]:  notice: Requesting fencing (reboot) of node node2
...
Apr  2 22:19:11 ip-10-0-0-17 fence_aws: Failed: Timed out waiting to power OFF
...
Apr  2 22:19:11 ip-10-0-0-17 stonith-ng[14285]:   error: Operation 'reboot' [18866] (call 42 from crmd.14289) for host 'node2' with device 'aws_fence' returned: -62 (Timer expired)
...
Apr  2 22:19:11 ip-10-0-0-17 crmd[14289]:  notice: Peer node2 was not terminated (reboot) by node1 on behalf of crmd.14289: Timer expired
...
Apr  2 22:19:11 ip-10-0-0-17 pengine[14288]: warning: Cluster node node2 will be fenced: peer is no longer part of the cluster
...
Apr  2 22:19:11 ip-10-0-0-17 crmd[14289]:  notice: Requesting fencing (reboot) of node node2
...
Apr  2 22:19:12 ip-10-0-0-17 fence_aws: Failed: Unable to obtain correct plug status or plug is not available
...
Apr  2 22:19:14 ip-10-0-0-17 fence_aws: Failed: Unable to obtain correct plug status or plug is not available
...
Apr  2 22:19:14 ip-10-0-0-17 stonith-ng[14285]:   error: Operation 'reboot' [19126] (call 43 from crmd.14289) for host 'node2' with device 'aws_fence' returned: -201 (Generic Pacemaker error)
Apr  2 22:19:14 ip-10-0-0-17 stonith-ng[14285]:  notice: Couldn't find anyone to fence (reboot) node2 with any device
Apr  2 22:19:14 ip-10-0-0-17 stonith-ng[14285]:   error: Operation reboot of node2 by <no-one> for crmd.14289@node1.5abdec11: No route to host