[Mageia-sysadm] [138] use a cname for puppet

Fri Nov 5 14:42:35 CET 2010

Le vendredi 05 novembre 2010 à 13:30 +0100, root at mageia.org a écrit :
+> Revision: 138
+> Author:   misc
+> Date:     2010-11-05 13:30:45 +0100 (Fri, 05 Nov 2010)
+> Log Message:
+> -----------
+> use a cname for puppet
+
+for the record, this broke puppet on all node, as noted by buchan :
+
+14:05:09|  blingme> info: Not using expired catalog for
+valstar.mageia.org from cache; expired at Fri Nov 05 14:00:46 +0100 2010
+14:20:43|  blingme> err: Could not retrieve catalog; skipping run
+14:21:12|  blingme> err: Could not retrieve catalog from remote server:
+hostname was not match with the server certificate
+
+the fix is explained here :
+
+http://projects.puppetlabs.com/projects/puppet/wiki/Ruby_Ssl_2007_006
+
+change the config on server side, 
+give the name of the vhost, 
+restart puppetmasterd,
+fix puppet on all nodes :
+   be sure that server = puppetmaster.mageia.org
+   restart puppet
+
+I plan to deploy puppet packages using puppet to avoid this next time
+( as this would also help to deploy report, and other nifty stuff ).
+
+But this cause some trouble with the puppetmaster, as it use the same
+config file than puppet agent ( /etc/puppet/puppet.conf ), so since the
+file on server is managed by svn ( ie, directly in the root ),
+the puppet.conf file must be in sync with the one deployed for the
+client ).
+
+I see 2 solutions :
+- keep in sync puppet.conf with modules/puppet/templates/puppet.conf 
+- (re)move puppet.conf so it doesn't conflict, and be sure that puppet
+deploy the proper one even on the master 
+
+First one is ugly and error prone ( but work as that's I use on my own
+server ), but the 2nd one is untested so maybe I have missed a obvious
+point. 
+
+WDYT ?
+
+-- 
+Michael Scherer
+
+