We're currently using rabbitmq in active/standby mode with drbd-backed storage. (I know it's not the current recommendation, may be able to change this eventually.)
Anyways, I'm trying to track down the expected behaviour if the rabbitmq server does a controlled switch of activity (which would involve a controlled shutdown of the server). I'm assuming that this will result in the server sending a "close connection" message for every open connection.
On the client side, is there any immediate indication of this happening? I'm thinking of callbacks being called, exceptions raised, etc.
For some background, I'm looking at OpenStack code that currently will wait for up to 60 seconds for an RPC call to time out (the client is waiting on a DirectConsumer queue for the RPC response). If the rabbitmq server does a controlled switchover during that interval, I'd like to detect this and raise an exception immediately instead of waiting for the full timeout.
Any pointers would be appreciated.