On 08/02/2008 11:17, Tom Hughes wrote: ... >> BUT, there appears to be a server setting for lighttpd >> 'max-keep-alive-requests' (default 128 - what is it on ours?) after >> which the connection is dropped. > > The config is in the repo: > > http://svn.openstreetmap.org/sites/rails_port/config/lighttpd.conf > > but we don't set that at all, so it is whatever the default is, which > according to the doco is 16 requests with a 30 second idle timeout.
Hmm. So it does. The first page I got Googling it was http://www.debianhelp.co.uk/performancelighttpd.htm which says it is 128. I guess I should believe lighttpd.net > The only thing is that allowing too much keep-alive might lead to > people managing to "hog" the server. Indeed. So what is causing this consistent upload failure at around 250 requests? Is this also something only I'm seeing in this completely consistent way, or does everyone get this? (Or do most people upload less more often so not get into the position in the first place?) David _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
