[Gluster-infra] POC ( Proof of concept ) for CM ?

Michael Scherer mscherer at redhat.com
Wed Sep 3 12:56:45 UTC 2014


so, in order to see if salt is suitable and to get a feeling of the
tools from everybody, and before we invest too much time on it, I would
propose we tackle a small CM project, ie manage just a part of the infra
with salt.

I usually divide the CM work in 2 types of tasks :
- applying configuration automatically ( ie, what puppet, chef, cfengine
do ), ie say "this is a webserver" and have the tool make it happen. 

- executing ad-hoc tasks ( where salt, func, ansible, capistrano,
mcollective shine ), ie saying "upgrade the software on all thoses boxes
1 by 1 with this set of tasks".

I would suggest we try to find 1 task for each, so we can have at least
a baseline.

For the configuration side, I guess it would be a good idea to take
something simple but not too much, would varnish be a good candidate ?

For the orchestration part, I would be tempted to do "upgrade", but
that's quite a common task to automate, so it may be too simple. The
idea is to have a task we run when needed, and see how we can setup a
automation for that. Something around jenkins might be a solution.

The idea will be to see how we can setup a repository, acl, etc, and the
general workflow, and then implement it, and iterate from here.

Does this approach look reasonable, too complex, etc ?
( if, so, I can start to propose a simple workflow for starting )

Michael Scherer
Open Source and Standards, Sysadmin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://www.gluster.org/pipermail/gluster-infra/attachments/20140903/6304c062/attachment.sig>

More information about the Gluster-infra mailing list