(This is already logged as ticket 247338, reposting here at the suggestion of SW tech support)
As an offshoot of problem ticket 226977, we determined that if an additional polling engine server starts up when the primary is also booting but is still not completely responsive, the additional poller fails to connect, (sometimes) fails to enumerate it's nodes/interfaces/volumes, and fails to collect SNMP statistics (although ping continues to work if it can find its nodes).
There is nothing to indicate this "partially up" state except the absence of SNMP statistics.
I'm requesting that the additional poller either:
1) retry to establish full connectivity in this situation
2) have the option of notifying the administrator if this happens, so that services can be restarted.
Good feedback, since you have it logged as a support ticket, they got it logged as an enhancement request into our system for me to look at for a future release.