Ed Loach wrote: > Once the current queue of "MissingTileRequest" tiles (and the occasional > missing bits of Alaska I notice in the queue) are done, will they ever need > doing again?
No. (And I really wish that AlaskaMissing guy with an IP address in Germany would stop whatever he is trying to do before I spent more effort in banning him). It is a one-off situation and that is why I am hesitant to spent enormous amounts of brain power of the problem. On the other hand it's going to be a one-off situation for quite a bit of time (ie months), as we are far from having finished the rerendering the of the world. > led to over 150,000 (when I started using [EMAIL PROTECTED], perhaps more > before then) bits of ocean needing rendering. Well, we are converting the old "legacy" tilesets to the new tileset file format by requesting a rerender. So there will be 16 million render requests until the world is finished... > If it is a bit of a one-off situation, then perhaps once the queue is empty > the full upload queue will be a fairly rare occurrence after that, and > worrying too much about the order the jobs are served to clients won't be > such an issue? Yes. Rather than investing enormous amounts of time in making the handout process smarter, that bottleneck could be totally eliminated if the client assembled the tileset file on the client side and uploaded that (kleptog mailed the code to do that to the list a while ago). This would go a far step in getting the work the upload processor needs to do close to 0. Also, if it's really a big bottleneck, I could start working in making the upload processor multi threaded. However, given that it copes just fine under "regular" load situations, I have other priorities ATM. spaetz _______________________________________________ Tilesathome mailing list [email protected] http://lists.openstreetmap.org/listinfo/tilesathome
