Yeah, Don't Run RabbitMQ on a Floppy Disk
If you observe that OpenStack Nova processes (Nova Compute in this case) are stuck while doing something like this:
2015-11-26 10:31:33.521 7750 INFO oslo.messaging._drivers.impl_rabbit [req-a66f13cc-4e26-43c6-98bd-fcd89164072c ] Connecting to AMQP server on localhost:5672
2015-11-26 10:31:33.545 7750 INFO oslo.messaging._drivers.impl_rabbit [req-a66f13cc-4e26-43c6-98bd-fcd89164072c ] Connected to AMQP server on localhost:5672
2015-11-26 10:31:33.550 7750 INFO oslo.messaging._drivers.impl_rabbit [req-a66f13cc-4e26-43c6-98bd-fcd89164072c ] Connecting to AMQP server on localhost:5672
2015-11-26 10:31:33.565 7750 INFO oslo.messaging._drivers.impl_rabbit [req-a66f13cc-4e26-43c6-98bd-fcd89164072c ] Connected to AMQP server on localhost:5672
And all the RabbitMQ connections seem to be blocked or blocking:
$ sudo rabbitmqctl list_connections
Listing connections ...
guest 127.0.0.1 41017 blocked
guest 127.0.0.1 41022 blocking
guest 127.0.0.1 41023 blocking
guest 127.0.0.1 41024 blocking
guest 127.0.0.1 41025 blocking
guest 127.0.0.1 41028 blocked
guest 127.0.0.1 41030 blocked
guest 127.0.0.1 41034 blocked
guest 127.0.0.1 41036 blocked
guest 127.0.0.1 41037 blocking
guest 127.0.0.1 41038 blocked
...done.
Then it probably means that you have almost ran out of hard disk space.
As always, I constantly forget this and start debugging Nova Compute until I think of listing RabbitMQ connections and try to find out why they are blocked. But the moment you clear enough disk space and the debug statements in the Nova process start to flow again is pretty enjoyable to watch.