Actions
Bug #19316
closedchacra nodes that are bounced don't persist the (correct) hostname - breaks rabbit+chacra
% Done:
0%
Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
After a few chacra nodes were turned off by OVH (and back on again) they came up like:
$ ssh 2.chacra.ceph.com ubuntu@2:~$
The problem (I believe) shows up because the hostname being used is a FQDN, so the short hostname translates into the first sub-domain (first split after the '.')
so for 3.chacra.ceph.com this means:
ubuntu@3:~$ hostname 3.chacra.ceph.com
Which breaks rabbitmq:
Mar 20 11:58:31 3.chacra.ceph.com rabbitmqctl[1930]: attempted to contact: [rabbit@3] Mar 20 11:58:31 3.chacra.ceph.com rabbitmqctl[1930]: rabbit@3: Mar 20 11:58:31 3.chacra.ceph.com rabbitmqctl[1930]: * unable to connect to epmd (port 4369) on 3: badarg (unknown POSIX error)
Because it sees '3' as being the hostname, which is wrong.
The /etc/hostname shows:
ubuntu@3:~$ cat /etc/hostname 3.chacra.ceph.com
And /etc/hosts has:
ubuntu@3:~$ cat /etc/hosts # Ansible managed: /Users/andrewschoen/dev/chacra/deploy/playbooks/roles/common/templates/hosts.j2 modified on 2016-07-19 09:02:00 by andrewschoen on Andrews-MacBook-Pro-2.local 127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4 ::1 localhost localhost.localdomain localhost6 localhost6.localdomain6 127.0.1.1 chacra3 158.69.93.173 3.chacra.ceph.com
After manually setting the hostname to `chacra3` (and 2.chacra.ceph.com to `chacra2`) rabbitmq was able to start.
Actions