epmd: node name already occupied epmd-starter-443584618

[René Schultz Madsen]

1. I have successful deployment (Ubuntu) with 3 controllers (node-11,12,15), 4 compute and 4 Ceph and 1 Elasticsearch Kibana, InfluxDB Grafana
2. Power off primary controller (node-11)
3. Wait 20 minutes
4. Start node-11

All services are running as expected, but LMA is filled with multiple lines of:
epmd: node name already occupied epmd-starter-443584618

I have tried to restart rabbitmq on all nodes, but the problem still persist.

PCS status: http://paste.openstack.org/show/487977/

Release 7.0 ISO

[René Schultz Madsen (p-rm-l)]

rabbitmq 1119 0.0 0.1 303784 18328 ? Ssl 12:33 0:00 /usr/lib/erlang/erts-5.10.4/bin/beam.smp -- -root /usr/lib/erlang -progname erl -- -home /var/lib/rabbitmq -- -sname epmd-starter-443584618 -noshell -eval halt().

Keeps coming up on all controllers.

[Bogdan Dobrelya (bogdando)]

Low, as it affects nothing

[Alexey Lebedeff (alebedev-a)]

The only job of epmd-starter-XXXX is to ensure that epmd is indeed running. Presence of this message just indicates that epmd is already running.

Fixed in rabbitmq 3.6.0 - https://github.com/rabbitmq/rabbitmq-common/pull/4

[Bogdan Dobrelya (bogdando)]

Thank you for a fast RCA, Alexey. Triaged as we have plans to switch to the rabbit 3.6.0 for the Mitaka

[René Schultz Madsen (p-rm-l)]

Issue was resolved manually by first killing epmd on node-11, then restarting the entire node, because rabbitmq refused to restart and, maybe related to https://bugs.launchpad.net/fuel/+bug/1525919

Buttonline was it can be resolved by restarting the controller node(s).

[Dmitry Pyzhov (dpyzhov)]

We passed SCF for 9.0. Moving the medium priority bug to 10.0 release.

[Michael Polenchuk (mpolenchuk)]

Marked it as "invalid" since 9.x already has {rabbit,"RabbitMQ","3.6.1"} on board.