Holy crap. I didn't even know that was a version? You haven't upgraded
since 2006? 2005?
Is your cluster currently exhibiting any issues or problems? Errors in
syslog/etc? Good to know about those up front before attempting an
upgrade.
I'd recommend you upgrade in stages. There's an upgrading guide on the
wiki... See if you can get up to version 2.0-something, run that for a
little bit, then try to go to 2.37 - or go all the way to 2.37.
Once you're on 2.37 and everything seems okay, I'd recommend running a
FSCK. That should ensure you're upraded from the old replication method
into the "new" table based one, and any files which got stuck over the
years are examined.
Let it run on 2.37 for a few days in case of any issues, then upgrade
straight from 2.37 to the latest version. The reason why I recommend 2.37
is that was where a lot of the architecture changed along with some DB
changes.
Once you're on the latest version, run FSCK again. The newer FSCK should
catch any final issues?