[PVE-User] Ansible role for Proxmox (4.x clusters)

Musee Ullah lae at lae.is
Fri Mar 10 20:37:58 CET 2017


On Wed, 2017-03-08 at 21:15 -0500, Jeff Palmer wrote:
> Musee,
> 
> I finally got a chance to test out this role.  I appreciate the
> effort. it's hugely helpful.
> 
> I have 3 suggestions to help improve the role.
> 
> #1) the ssh port should be a confirable variable.  a lot of people
> don't run ssh on 22. (this is an opinion, and as such, completely
> optional)
> 
> #2) Proxmox recommends removing the os-prober package to reduce the
> possibility of corruption in vms and containers:
>     See https://pve.proxmox.com/wiki/Install_Proxmox_VE_on_Debian_Jes
> sie#Recommended:_remove_the_os-prober_package
> for reference.
> 
> #3) Each node should have all other nodes listed in the /etc/hosts
> files.  Currently, it only adds the local hostname to it.
>       Additionally,  the local node should add "pvelocalhost" to the
> /etc/hosts entry.  An example for a host of "10.0.0.2"
> 
> 10.0.0.1 mox1.example.com mox1
> 10.0.0.2 mox2.example.com mox2 pvelocalhost
> 10.0.0.3 mox3.example.com mox3
> ...
> 
> 
> I did notice that if I tried changing the cluster name (in this case,
> mox-dfw)  I got several errors.   I didn't get a chance to try to
> diagnose/debug why it failed.  When I removed that option, and used
> the defauls, all was fine.
> 
> 
> Overall, I'd say this is great work.  Thanks for your efforts! Hope
> this feedback can help.

(reposting to list)

Hi Jeff,

Thanks for the feedback!

1) This is probably out of scope for this Proxmox role - I think it
should probably be left up to the user to have e.g. a separate role or
task for configuring SSH for machines in the cluster. The way this role
updates the SSHD configuration isn't very invasive, using blockinfile
instead of templating the entire SSHD configuration.

2) I opened up an issue for this - https://github.com/lae/ansible-role-
proxmox/issues/10

3) Also opened up an issue for this - https://github.com/lae/ansible-ro
le-proxmox/issues/11. I think I did consider this before, but got stuck
on deciding how to deal with cases where the hostname on the machine
and inventory don't match up.  I should probably just trust the
hostname on the machine, I suppose.

What do you mean by changing the cluster name? Are you talking about
the "pve_group" variable?

lae

P.S. Thanks Patrik for letting me know of the JS location change.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: This is a digitally signed message part
URL: <http://pve.proxmox.com/pipermail/pve-user/attachments/20170310/7dc801c3/attachment.sig>


More information about the pve-user mailing list