|
From: | David Decotigny |
Subject: | [Maposmatic-dev] Re: [Maposmatic-contact] 6037 onwards: Rendering failed, please contact address@hidden |
Date: | Thu, 07 Jan 2010 09:33:02 +0100 |
User-agent: | Thunderbird 2.0.0.23 (X11/20090817) |
Hello Planet, Ian Mellor wrote:
Hi Maposmatic!In case nobody has told you, although I suspect someone will have mentioned it...the last map to be successfully rendered was http://www.maposmatic.org/jobs/6036 at 06:40 this morning. I can't find any maps after then that have actually rendered.Ian
Yes, we've had a few hickups with our sql database: once one rendering job took too long (> 20mn) to perform its sql queries, then 1/ the rendering failed and got aborted, 2/ but not its sql backend which kept on running (and eating 100% CPU), so 3/ chances were very high that the next rendering jobs timed out, which caused the next to time out, etc. This is a perfect example of a domino effect.
As per https://savannah.nongnu.org/task/?10040 this should be worked around since yesterday night. It doesn't prevent occasional jobs to time out (when the rendering area is large and/or dense or the DB very loaded), but it should prevent the jobs /after/ the timeout to be OK (=> no domino effect). We still need to optimize our requests to prevent the timeout from happening altogether.
Regards,
[Prev in Thread] | Current Thread | [Next in Thread] |