[Tilesathome] timeouts and retries (changes)

[spaetz]

I was a little unsatisfied with our current request timeout situation. We could either delete or retry requests...
I implemented a max_retry feature (a couple of lines in timeout.php). Starting immediately the request queue is handled as follows:

- After 24 hours in the active queue (and no upload finishing), the request is moved to the new queue again and handed out to another renderer.
- If after another 25h no upload happended we delete the active request. If one of the laggards still wants to upload his results later (I think) it may still do but it's not in the active queue anymore.

The number of retries and timeouts is configurable so we may want to tune them a bit according to what renderers experience.

Let's see if this works out in practice, I haven't tested it yet in production.

The timeout script used to delete/move all requests individually, it does now with only one query. Because of this I changed ojw's crontab to run this script once per hour rather than 2-hourly. This should give us less of a saw effect in the stats.

spaetz
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 186 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/tilesathome/attachments/20070727/13bc603f/attachment.pgp>