High Availability cluster shows non-existent estranged nodes in clustat output, or has "Received message has invalid digest... ignoring" messages in logs

Issue

• A cluster node has many messages like the following in /var/log/messages:

  Jan 24 04:08:25 node1 openais[743]: [MAIN ] Received message has invalid digest... ignoring.  
  Jan 24 04:08:25 node1 openais[743]: [MAIN ] Invalid packet data
  - OR -
  Jan 24 04:08:25 node1 corosync[743]: [MAIN ] Received message has invalid digest... ignoring.  
  Jan 24 04:08:25 node1 corosync[743]: [MAIN ] Invalid packet data
  

• clustat output shows estranged nodes in the node list that aren't a part of the cluster:

  Cluster Status for test-cluster @ Wed Oct  7 10:18:38 2009
  Member Status: Quorate
  
   Member Name                             ID   Status
   ------ ----                             ---- ------
   node1.test                                  2 Online, Local
   node2.test                                  1 Online
   node7.prod                                  7 Offline, Estranged
  ~~~
  

• Cluster shows odd behavior such as pulling the wrong configuration file or config_version number when using cman_tool version -r or other synchronization mechanism.