Comment 5 for bug 1579128

Revision history for this message
Ilya Shakhat (shakhat) wrote : Re: Lost communication between nova-conductor and nova-compute

Investigation shows 2 issues:

1) The reply queue exists in RabbitMQ, but the corresponding exchange is not. The following error is presented in RabbitMQ log:

=ERROR REPORT==== 3-May-2016::21:36:02 ===
Channel error on connection <0.17745.2> (192.168.0.126:55899 -> 192.168.0.131:5673, vhost: '/', user: 'nova'), channel 1:
operation queue.declare caused a channel exception not_found: "failed to perform operation on queue 'reply_09181cdbdf894afeabcc99be8983f528' in vhost '/' due to timeout"

As result all messages published to the queue fail with:

=ERROR REPORT==== 3-May-2016::22:43:21 ===
Channel error on connection <0.9232.15> (192.168.0.131:51606 -> 192.168.0.131:5673, vhost: '/', user: 'nova'), channel 1:
operation basic.publish caused a channel exception not_found: "no exchange 'reply_09181cdbdf894afeabcc99be8983f528' in vhost '/'"

2) All threads inside Nova-conductor stall in read forever, as result the service cannot process messages nor reply to them. A separate bug will be filed to track this.