Bug #13679
closedceph-cm-ansible: correct 'cloud.front' hostnames
0%
Description
See #13365
Instead of the "Correct hostname if it is 'localhost'" step in the testnodes role, we should set the hostname we know a node should have.
Updated by Zack Cerza over 8 years ago
- Status changed from New to Fix Under Review
- Assignee set to Zack Cerza
Updated by Zack Cerza over 8 years ago
A bit of background:
When I investigated what was happening in #13365, I initially wasn't reproducing the problem. CentOS 7 instances created with downburst were correctly getting their real hostname. Then I noticed the logs showed us using an upstream kernel, which would require a reboot after installation. I began to suspect cloud-init
.
Sure enough, after rebooting my downburst instances, the hostname was set to 'cloud.front.sepia.ceph.com'. Turns out cloud-init
isn't writing to /etc/hostname
and instead only using /bin/hostname
.
The PR drops the fancy logic and uses the ansible hostname module unconditionally. That module is idempotent and its effects are permanent.
Updated by Yuri Weinstein over 8 years ago
- Has duplicate Bug #13463: "Failure: ceph-deploy was not able to gatherkeys" on centos 7.0 added
Updated by Zack Cerza over 8 years ago
- Status changed from Fix Under Review to Resolved