==PLEASE NOTE: THESE INSTRUCTIONS ARE NOW OUTDATED, SEE THIS POST FOR AN UPDATE ==
After I dabbled with CentOS 7 a few weeks ago I decided to take the plunge and upgrade the few servers I run at DigitalOcean to CentOS 7. I run two machines, one is a Puppet and Salt Master, the second runs Docker.
Puppet Master Server
A while back I added a few bash scripts to deploy both a Puppet Master & Agent to GitHub, as Puppet provide a repo for CentOS 7 I decided to just update the installation script to use this repo and then give ago. I use stephenrjohnson/puppet to deploy the Puppet Master. the Puppet part worked, however it failed to install Passenger and its Apache module because there is no mod_passenger available for CentOS 7 yet. To get around this I grabbed the require packages from The Foremans nightly builds repo. This turned the initial yum install line into the following;
[code gutter=”false”] yum install -y vim-enhanced git http://yum.puppetlabs.com/puppetlabs-release-el-7.noarch.rpm https://anorien.csc.warwick.ac.uk/mirrors/epel/beta/7/x86_64/epel-release-7-0.2.noarch.rpm http://yum.theforeman.org/nightly/el7/x86_64/mod_passenger-4.0.18-9.5.el7.x86_64.rpm http://yum.theforeman.org/nightly/el7/x86_64/rubygem-passenger-native-4.0.18-9.5.el7.x86_64.rpm http://yum.theforeman.org/nightly/el7/x86_64/rubygem-passenger-native-libs-4.0.18-9.5.el7.x86_64.rpm http://yum.theforeman.org/nightly/el7/x86_64/rubygem-passenger-4.0.18-9.5.el7.x86_64.rpm http://yum.theforeman.org/nightly/el7/x86_64/rubygem-rack-1.4.5-3.el7.noarch.rpm http://yum.theforeman.org/nightly/el7/x86_64/rubygem-rack-doc-1.4.5-3.el7.noarch.rpm [/code]
Now everything was being installed correctly, however Apache refused to start due to the following error;
[code gutter=”false”] Invalid command ‘RackAutoDetect’, perhaps misspelled or defined by a module not included in the server configuration [/code]
This was due to the presence of RackAutoDetect Off and RailsAutoDetect Off in the puppet_passenger.conf.erb file. Once these two lines were removed everything worked.
The final script can be found here
I have had my base Puppet Configration on GitHub for a while. So my next move was to grab a copy of the configuration and try it. For the most part this worked, there were a few errors though.
- man was not available so was removed
- jwhois was replaced with whois
- The firewall module was not able to use iptables-save even though iptables was installed and configured, installing iptables-services resolved this
The rest of the packages and configuration worked as expected.
This time I tried the Puppet Agent install script, it worked first time;
[code gutter=”false”] curl -fsS https://raw2.github.com/russmckendrick/puppet-install/master/agent-el7 | bash -s puppet.master.com [/code]
Once I had Puppet installed I ran puppet agent — test and was greeted with lots of NGINX errors, but thats all. Everything else worked as expected.
The only issue here was the Puppet module I am using to install and configure NGINX was using http://nginx.org/packages/rhel/6/ I updated it to use http://nginx.org/packages/centos/7/ and hey presto, server is back to working as before.
In all, other than having to tweak some packages to get services running it wasn’t as bad as I thought it would be. Considering the release is still less than a month old I expected to have to do a lot more work than I did to get everything up and running how I had it previously.
So would I run it production?
Yes & No. If I was running a service which didn’t anything other standard repository packages then I would have no problem running it in production, for some of the none standard stuff like the Puppet Master server, I would give it a few more weeks until the required packages start to filter down to more production ready repositories.