How to Handle RabbitMQ peak after restarting neutron-openvswitch-agent

MaSch themasch at gmx.net
Tue Mar 9 07:57:19 UTC 2021


Hello at all.

After adding an new direct attached VLAN to our computeNodes and
restarting neutron-openvswitch-agent, we detected a huge peak in
RabbitMQ Messages.
Mostly "neutron-vo-Port-1.1_fanout" these messages seem every node
produced  5000+ Messages.

Due to a bridge_mapping mismatch, the agent repeatedly restarted and
produced more and more Messages.
At the peak there were about 2Million messages in the queue.
As we ran into Network issues caused by Messaging timeouts i would like
to know if there is a procedure to handle this messages.
Is it save to delete the queue as those messages?

It seems they disappeared after a timeout period of about 30 Minutes.

are currently using
OpenStack Queens release.

Thanks a lot in advance.


Best regards
MaSch




More information about the openstack-discuss mailing list