Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have 2 crucial servers that are running an unsupported & ancient version of Linux, and there's currently no disaster plan for them.
I want to set up a backup system for them, and be able to create a foolprof disaster plan.

The servers cannot be down, so freezing them for snapshots, or replication is out of the question. I looked at Chef and puppet, i don't know if those are worth a shot. I would like to hear recommendations.

share|improve this question
1  
+1 for knowing enough to know you're in a bad place and wanting to get out of it, but be aware that "a foolproof disaster plan" is intensely site-specific -- you probably won't get that from some answers on Server Fault, you have a long requirements analysis and system design project ahead of you my friend... –  voretaq7 Sep 5 '12 at 22:53

1 Answer 1

up vote 5 down vote accepted

You are about a million miles away from being ready to consider Puppet or Chef.
Worry about your (nonexistent) backups and disaster recovery plan first.

Were I in your shoes (I was 3 years ago), this is what I would do --


Step 0: Get some kind of backup. NOW.
If these machines are really mission critical just plug in a big USB hard drive and grab a tar archive of the whole system (see man tar on your system).
Once you at least have an archive, try to locate the original install media for the system, spin up a virtual machine (use Virtualbox or something similar) and do a clean install of your ancient Linux version.

Now figure out how you would restore to a working state if a meteor should strike and take out your production machines.


STOP
Go no further until you have completed step zero and can restore to a working configuration in the event of a disaster. The Chaos Monkey is watching you.


Step 1: Select, install and configure a backup system.

My recommendation for you would be Bacula. Get it working and get regular backups going.
Perform restore tests to your virtual machine from Step 0 and make sure you can get from a bare clean install back to a working system.

Perform this test frequently from now on. (I do it every 3 months. Seriously.)


Step 2: Get on to a supported release of your operating system.

Take that virtual machine from step zero and test upgrade paths until you find one that works.
Best case? Nice clean easy vendor-supported upgrade.
Worst case? Reinstall and some manual pain and suffering.

While you're doing this you can start looking at tools like Puppet and Chef and thinking about how you want to implement configuration management.

(Don't forget those backups and restore tests once you do the upgrade.


Step 3:* Configuration Management

Now that you have an environment that's not in danger of blowing up you can breathe easy and consider configuration management tools in a calm, planned manner.

You can roll this into step 2 and make it part and parcel of your OS upgrade, or you can stabilize everything manually and then bring in configuration management later -- that's your choice.

When you're ready for this step come back and take a long look through the and questions. You may also find interesting information tucked away under

share|improve this answer
    
Note that you're allowed to mix and match steps here: If your installation is so hopeless that you'll never get bacula running go ahead and keep doing tarball backups until you can replace it. What's important is that you get to a point where you can go from "It's all gone" to "It's all back" with confidence (and it's better if you get there fast). –  voretaq7 Sep 5 '12 at 22:59
    
Kickass anwser! –  Jorge Sanchez Sep 5 '12 at 23:31
    
Again, repeat, Kickass Anwser –  Jorge Sanchez Sep 6 '12 at 0:06

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.