I was just reading Brad's post about what happened yesterday, and I started thinking about something that might have the potential to help alleviate some of his stress....I hope so, at least.
Most service providers have a link somewhere on their home pages that shows server status with messages about any problems occurring at the time. I know that lj_maintenance is there for such things, but most users don't use it or know it's there, it seems. Plus, the messages usually show up after the problem was resolved.
So my idea was that somewhere in the tab we could add a link for Server Status where someone could post a quick message when there were issues. It'd also be neat to show stats for the load and maybe show them for the last month to show users when the peak times are. If all the stats stuff isn't needed, then maybe we could just put the server status link there and link that to lj_maintenance, so more users would see that lj_maintenance is there. And then, maybe a quick message could be posted as soon as a problem is recognized.
I am open to any feedback. If this type of thing already exists, then I apologize. It was just a brain storm I had, and I'd really like to help alleviate stress for Brad and the other people working hard on making LJ so incredibly great!
So, it turns out that yesterday's slowness was caused by another infinite loop in the code. This happened a few months ago and it took quite awhile to track down.
So, uh... how's that auditing going, eh?
Whoever finds it gets a permanent account ... and if you already have one, you can give it to a friend. :-P
I wish the damn perl debugger would let me attach to a running process. Grrrrr.