retrace* machines managed by ansible

Kevin Fenzi kevin at scrye.com
Wed Apr 29 23:53:58 UTC 2015


On Wed, 29 Apr 2015 14:21:27 +0200
Richard Marko <rmarko at fedoraproject.org> wrote:

> On 04/28/2015 07:23 PM, Kevin Fenzi wrote:
> > On Tue, 28 Apr 2015 16:07:46 +0200
> > Richard Marko <rmarko at fedoraproject.org> wrote:
> >
> >> Hi!
> >>
> >> I'm working on https://github.com/abrt/abrt-ansible for the
> >> purpose of completely managing retrace* machines with ansible.
> > ok. Cool. 
> >
> > Would you be open to just merging this into our infrastructure
> > ansible? 
> 
> Yup but I would still like to maintain a current repository for
> stand-alone deployments.

Very reasonable. 

> >> Is anyone willing to take a brief look at mentioned recipes and
> >> point me in the right direction? In particular, what can/should be
> >> reused from fedora infra ansible and what should be improved in
> >> current recipes.
> > I can try and do so at some point. It's kind of hard to review this
> > sort of thing detached from real machines/setup. ;) 
> 
> I think we only need two faf specific roles added to fedora infra -
> storage & web that represents backend and frontend of the server. I
> will move this to roles/faf/(storage|web) directory so it's clear
> that these are specific for faf.

ok. 

> We can re-use httpd, memcached and db roles. Our db role is basically
> a postgresql with pg-semver extension loaded.

We could add that into our postgresql role as a conditional ?
 
> I'll also try tagging recipes similar to what fedora infra uses.

Sounds good. 

> I'm not sure about one thing which is the first time setup - I'm using
> first_time_setup variable for this that will initialize postgres
> database, creates tables required by faf and populates them with some
> data. This probably isn't the right way to do it but it's only
> required once so it's good enough for me now.

There's some tricks to this... one is to use a 'creates=/path/to/db'
and so it only runs when that db isn't there. Look at the mailman roles
in our repo... 

Or, we could split this out into a 'manual' playbook. (ie, in
playbooks/manual), which are playbooks only run manually for specific
things. 

kevin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://lists.fedoraproject.org/pipermail/infrastructure/attachments/20150429/f0b1f159/attachment.sig>


More information about the infrastructure mailing list