Customer needs help restoring system after network disruption. He has written "audit" code which uses the admin api. It will run on a scheduled basis to detect and correct issues caused by network disruption, split brain and split brain recovery.
Inconsistencies causes by split brain include (but not limited to):
puInstance in stopped state
2 gsm are active managers for same pu.
admin api reflects different number of pu than gsm.
esm in looping restart because admin api showing different pu counts than gsm.
This JIRA is associated with following Salesforce issue.
salesforce case 12325 - Multiple PUs stay in COMPROMISED state after a network glitch
Client is requesting working session for help with this.