Oldie but goodie by Nelson Elhage at the “Made of Bugs” blog. I mostly agree with most of these points, and consistently break the rule about only keeping a chronological journal, as opposed to trying to bucket them under application-specific or domain-specific journals. (That’s a flavor of Premature Optimization and don’t try to let anyone tell you otherwise. [And right here you can expect me to insert all of the exceptions to that rule that I’ve cooked up over the years.])
One thing that should go without saying, and is a bit of advice I’ve given over and over again that dovetails well with Elhage’s post here: feel free to slow down on your “lab notebook” at any time, but the moment that you feel yourself getting scattered, go right back to a disciplined notebook. Nothing makes you slow down and focus on The Right Things quite like keeping a record — no matter how informal that record it.