I use Puppet for my complex servers, but my Ceph machines are lightweight, and Puppet, for all its virtues does require a Puppet agent to be installed on each target and have a corresponding node manifest.

For the Ceph machines I just use Ansible. Since all my /etc/ceph files are identical on all machines, I can define a set of hosts to Ansible as belonging to a "ceph" group, and add/remove as needed, then use Ansible to update them /en masse/. Ansible requires no additional software on its targets to act as agents, as it simply directs operations via ssh.

I don't have special /etc/ceph files for my rgw hosts, as since at least Octopus on, the pertinent information is in Ceph's configuration database and the actual contents of /etc/ceph are minimal. Ansible simply clones its master copy of /etc/ceph to each ceph target.

On 6/10/25 07:15, Albert Shih wrote:
Le 10/06/2025 à 10:02:40+0200, Boris a écrit
Him

you can use puppet to set ceph conf commands.
   # Disable RadosGW access logs
   exec { 'set-debug_rgw_access':
     command => 'ceph config set client.rgw debug_rgw_access 0',
     unless  => 'ceph config get client.rgw debug_rgw_access | grep -w 0',
     path    => ['/usr/bin', '/bin'],
     require => Package['ceph-common'],
   }
Ok. I already do that for other thing. I just didn't know if I can do that
(vs put everything in ceph.conf).

I also managed the ceph.conf a long time with puppet. It is just an ini
file.

If you use the ceph orchestrator, you can define most, if not all,
parameters in the service definition:
https://docs.ceph.com/en/quincy/cephadm/services/rgw/
OK good to know.

Another thing, that comes to my mind is to generate a dedicated config for
the rgw daemon and start the rgw daemon with the dedicated ceph config
file. Maybe you can also set CLI parameter, but I never looked into that.
I have big preference for the ceph config set method. So I would try that.

Thanks.

Regards
_______________________________________________
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io

Reply via email to