Nothing to suggest that it has crashed (we see those too often also but generally the two aren't linked). When it occurs all processes are running ok and it's often not for every user that the issue occurs but generally it's either for just one or two or sometimes everybody at once. I had thought that it would be a session…

That's very much my experience with SRs ... getting the right person early on helps a lot, otherwise it's a VERY slow and frustrating process ..... we have some that have been open for over a year at Sev 1!

That's unfortunately the conclusion I've come to also. Have raised and SR but received no real answer either way so that's why I came here :-) Thanks for your input anyway.

We are using a shared directory for cache, RPD and catalog (as per standard cluster setup)

Yes, the only issue is that we cannot do an RPD deployment (either offline or online) due to the master server being offline. Was easy to define the master server in 11g but seems to be no obvious way within 12c.

We're running in an HA environment (using Oracle's recommended method) so can start the Admin Server on any of the servers in the cluster via a share. As server 1 is unavailable we're running the AS on server 2 but it still thinks that server 1 is the master - even though it's offline.