I think it would be great to have a high level troubleshooting guide in the Manual that describes what we do to address common problems or errors. Rather than address specific questions about how to deal with specific problems that may arise, I want this document to explain:

• Where things are logged. Where an administrator/operator should look for information.
• Leading causes of common problems (replication lag, uneven chunk distribution, spurious failovers, etc.) as well as...
• Differential diagnosis patterns for common issues.

This may involve a rewrite or reorganization of the content in the monitoring document in the manual (i.e. <http://docs.mongodb.org/manual/administration/monitoring/>)

I think the best way to proceed with this would be to workout an outline and then make sure we're on the same page and then go from there. I'm much more interested in having all of the right information and the right suggestions than I am in having all of the right words.