The Linode Library has proved invaluable lately. I believe they may have one of the most comprehensive solutions for properly installing MongoDB out there. The best part, to me, is that they have separate walkthroughs for Debian, Fedora, CentOS, and Ubuntu. I would highly recommend checking out these guides when you prepare for your next (or first) MongoDB install. They not only cover the actual install, but they cover setting up bash scripts, adding Mongo to your system startup, and other useful tidbits.

• Install and Setup MongoDB on Debian 5 (Lenny)
• Install and Setup MongoDB on Ubuntu 9.10 (Karmic)
• Install and Setup MongoDB on Ubuntu 10.04 (Lucid)
• Install and Setup MongoDB on CentOS 5
• Install and Setup MongoDB on Fedora 13
• Install and Setup MongoDB on Fedora 12
• Build Database Clusters with MongoDB

A word of warning; The Database Clusters guide is slightly dated with the new release of Mongo 1.6. The documentation contains information for setting up Replica Pairs and not Replica Sets. The document is, however, a good source of information for setting up quick master/slave replication with 2 nodes.

Happy installation, it’s really quite painless!

I have posted a couple of important updates to MongoSession (a PHP MongoDB Session Handler) that I’d like to share with you.

1. I have removed the secondary indexed id field which should speed up operations as well as reduce the storage requirements. This is most beneficial for larger sites supporting a high rate of concurrent users.
2. I recently added two boolean constants, FSYNC and SAFE. These two constants are documented on the MongoCollection::update() page and can be turned on to ensure your session data is actually written to the server. Please note that turning these variables on will cause a performance hit to the overall session handling.
3. Both the garbage collection and session write methods have been updated to perform atomic operations. This was done as a measure to reduce the risk of race conditions as mentioned on the project page.
4. A bug was fixed in the read method where an extra line of code existed for no reason.
5. A bug in the garbage collector was fixed. Update previously only updated a single result as opposed to the whole set of matches.

Click here to be taken to the project page.

Questions, comments, forks, and bugfixes are welcome.