Second week of work...
Oct. 25th, 2003 03:08 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
johnpalmerSo, this second week, my boss (Misha) and I were going to Denver to meet with the development team. I've been studying SQL and running some traces on the database to get some ideas about what might need to be done.
The air that's on airplanes is *dry*, and I don't like it. Well, the air in Denver is *also* dry! My nose is not very happy with me right now.
But... we met with the developers, and they're good folks; they're the kind I can deal with, work with, and talk to. I spent extra time talking to the guy who'd done database administration prior to this, because I don't want him to think I'm there to steal his job. I hope that we'll work well together, because our jobs *are* separate, though linked.
We went out to lunch, and we chatted, and things were going okay... and then I came back. IIS was crashing due to database lockups; the database was locking up due to deadlocking. This is when one process has Resource A locked, and won't let go until it does the next thing, which requires Resource B; the other process has Process B locked, and won't let go until it does the next thing, which requires Resource A.
(It can be more complicated than that (there can be chains of locks that have the same effect), but that's the basic idea.)
By chance, I'd noticed that certain statements were taking a huge amount of time to process, and this can help cause deadlocking. I also noticed that the table that was at the center of the problem didn't have an index on an important column. So I indexed it, right then and there, and poof... the deadlocking problem was fixed (and, as near as I can tell, the long running statements were now running some 5 - 10 times as fast!)
I spent that night at a bar, drinking Guiness on tap and chatting around a pool table with three of the developers, and I think things went well. The next day, to set things in cement, I brought in cookies :-).
Things *could* go south, still, but I think we've got a good, solid working foundation on a new relationship, and I think things went amazingly well.
The air that's on airplanes is *dry*, and I don't like it. Well, the air in Denver is *also* dry! My nose is not very happy with me right now.
But... we met with the developers, and they're good folks; they're the kind I can deal with, work with, and talk to. I spent extra time talking to the guy who'd done database administration prior to this, because I don't want him to think I'm there to steal his job. I hope that we'll work well together, because our jobs *are* separate, though linked.
We went out to lunch, and we chatted, and things were going okay... and then I came back. IIS was crashing due to database lockups; the database was locking up due to deadlocking. This is when one process has Resource A locked, and won't let go until it does the next thing, which requires Resource B; the other process has Process B locked, and won't let go until it does the next thing, which requires Resource A.
(It can be more complicated than that (there can be chains of locks that have the same effect), but that's the basic idea.)
By chance, I'd noticed that certain statements were taking a huge amount of time to process, and this can help cause deadlocking. I also noticed that the table that was at the center of the problem didn't have an index on an important column. So I indexed it, right then and there, and poof... the deadlocking problem was fixed (and, as near as I can tell, the long running statements were now running some 5 - 10 times as fast!)
I spent that night at a bar, drinking Guiness on tap and chatting around a pool table with three of the developers, and I think things went well. The next day, to set things in cement, I brought in cookies :-).
Things *could* go south, still, but I think we've got a good, solid working foundation on a new relationship, and I think things went amazingly well.
