<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">Yeah, not a bad idea. Or I could just set up self hosted puppetmaster on all nodes and just rsync or git-deploy the changes from one to the others. Either way.<div><br></div><div><blockquote type="cite"><div bgcolor="#FFFFFF" text="#000000">It may be possible, but would likely be quite a mess. </div></blockquote>Why would it be a mess? It would only be a one time config change on the puppet agents (as well as signing the puppet certs on the puppetmaster).</div><div><br></div><div><div><blockquote type="cite"><div bgcolor="#FFFFFF" text="#000000">Ordinarily you would develop your changes on a single instance and get them reviewed and merged before applying to other instances.</div></blockquote>In this case, no. Kraken is a cluster setup (Hadoop, Kafka, Zookeeper, Storm, etc.), and the puppetization will be developed in labs. I need to test the changes somewhere before they are reviewed and merged.</div><div><br></div><div><br></div><div><br><div><div>On Apr 3, 2013, at 11:26 AM, Andrew Bogott <<a href="mailto:abogott@wikimedia.org">abogott@wikimedia.org</a>> wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite">
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
<div bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">On 4/3/13 10:16 AM, Andrew Otto wrote:<br>
</div>
<blockquote cite="mid:6B3E9AF5-8A3A-4356-9CF7-A11163A6DA20@wikimedia.org" type="cite">
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<div>Hi yalls,<br>
<font color="#0f61c8"><br>
</font>Analytics will soon be working on a kraken labs project.
We'd like to have a few nodes there, probably starting with 5,
but possibly up to 10ish.<br>
<font color="#0f61c8"><br>
</font>Is it possible/allowed to set up self-hosted puppetmaster
on a single one of these nodes, and have the other puppet
clients point at it?</div>
</blockquote>
<br>
It may be possible, but would likely be quite a mess. Ordinarily
you would develop your changes on a single instance and get them
reviewed and merged before applying to other instances.<br>
<br>
If there's a complicated reason why that won't work (for example,
you need to test them on multiple nodes before you regard them as
correct enough to merge) then here's the hacky method I would
recommend:<br>
<br>
1) Install puppetmaster::self on all nodes<br>
2) Commit your in-testing patch to gerrit, but mark it as a 'work
in progress' in review so that reviewers know to ignore it<br>
3) Use the cherry-pick gizmo in gerrit to cherry-pick the patch
onto other nodes<br>
4) test<br>
5) go to step 2<br>
<br>
I use that process pretty frequently; it's not as big of a pain as
it sounds. Ryan can probably set you up with a spiffy one-step salt
command to perform step 3 on many nodes at once. I don't know the
syntax for this offhand.<br>
<br>
-Andrew<br>
<br>
</div>
</blockquote></div><br></div></div></body></html>