Sunday, November 27, 2011

Using Twitter To Monitor Your Linux Server!

Yes, you can use Twitter for monitoring your server. I won't say that it is a complete monitoring solution nor I will ask you to throw away your existing monitoring mechanisms. In fact the script I am talking about, MYST (AGPLv3), was not created for this purpose. It was created so that I could tweet without using browser with inspiration from Hiemanshu, a Fedora contributor, using python-twitter api.

It is just a fun script which you can use to tweet the health of your server periodically to a private account which only a moderated set of people can follow. So here is how you do it:

Step 1: Create a Twitter account. From settings page, mark it private.

Step 2: Open Twitter's new application page and fill the form. Put the name as 'MYST' and website as ''.

Step 3: Download 'MYST: Twitter for Shell' and extract it.

Step 4: Open your application 'MYST' listed at Twitter's apps page and fill .myst.conf with the relevant details.
Copy it to your home directory.

Step 5: Install python-twitter (version 0.82) on your server along with dependencies.

Step 6: Put a cron with appropriate time (and path to scripts) to execute the following periodically:

./ update `./`

This is a cool method to check out the system health and you can modify to add more parameters to monitor.
Please contribute to the MYST project on

Wednesday, November 2, 2011

How To Install And Configure Puppet (Getting Puppet Up)

Puppet is a system management tool used my many large and small enterprises to manage their infrastructure. From the top of my head Twitter, Wikipedia, Digg, Nokia, Rackspace are some of the companies using it and there is no reason that you cannot use it to manage that single server you love or the entire data center you own.

Installing puppet is not difficult but i"ll recommend installing puppet on Fedora/EL from tmz repo instead of the official repo. Official repo is generally out of date while tmz repo has latest builds. I don't know if some such repo exists for debian-like distributions. If you are interested in installing from source then you should check out this page. Puppet follows client - server architecture. So you need to run a server and a client (which can be on the same machine). Install "puppet" on the client side and "puppet-server" on the server side.

Now let us start with building the server:

Puppet Server
Step 1: Configure tmz repo and install puppet-server

yum install puppet-server

Step 2: Puppet, by default, looks for site.pp file in manifest directory. Let us create one, if not present already.
mkdir /etc/puppet/manifests
touch /etc/puppet/manifests/site.pp

Step 3: Start the puppet master (a.k.a. server) using:
service puppetmaster start

Puppet Client
Step 1: Tell the client where is the server by adding server entry in [main] section:

Step 2: Start the puppet client
service puppet start

Puppet client will request a certificate from master. Now let us go to the master and sign the certificate.
You can check out all the signing requests by firing following on puppet master:
puppet cert --list

Sign the correct certificate by:
puppet cert --sign

Our puppet is up and running and ready to use. I'll build some manifests and modules to manage applications in next post or if you want, you can catch me at Fedora Users and Developers Conference at Pune, India on 6 Novemeber, 2011 where I'll build some manifests and modules live as a part of hackfest event.