On Thu, 2009-03-26 at 14:43 +0000, Andy Allan wrote:
> On Thu, Mar 26, 2009 at 12:00 PM, Shaun McDonald
> <[email protected]> wrote:
> > This is the weekly re-import of the data, when the render daemon is
> > stopped for the duration.
>
> I'm not entirely sure that it is, especially since I'd have expected
> it to restart yesterday morning (or maybe this morning - again, not
> sure when Jon does things). But looking at the munin graphs [1] I can
> see that the eth0 load is high, the CPU has been flat out earlier
> today, and the load average is spiking pretty high too.
There was a spike in the number of Apache threads and render daemon
crashed when it used up all 1024 available file descriptors. I have
known about this for a while, the render daemon has hit this a couple of
times over the past year.
I have restarted the process with the ulimit on the descriptors raised
to 4096. Longer term I think I either need rethink how we handle >1k
Apache connections or make the process increase the descriptors itself.
Jon
_______________________________________________
talk mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/talk
