- From: Ville Skyttä <ville.skytta@iki.fi>
- Date: Thu, 09 Oct 2003 21:24:53 +0300
- To: Olivier Thereaux <ot@w3.org>
- Cc: public-qa-dev@w3.org
On Thu, 2003-10-09 at 19:46, Olivier Thereaux wrote: > I just got reports of "sluggishness" for checklink. I checked, and it > seems indeed like checklink takes > 5 seconds to HEAD each resource. [...] > > Ville, do you have any idea what could cause this, or any experience of > similar trouble? At least there are some DNS problems on v.w.o. The primary nameserver (ie the first one in /etc/resolv.conf) doesn't answer at all, others seem to be fine. The effect can be witnessed eg. by running the following several times and comparing the results: time HEAD -d http://www.w3.org/ time HEAD -d http://18.29.1.34/ A 5 second DNS timeout somewhere? I was about to suggest running nscd on v.w.o but it seems to be running named anyway... There are some problems with checklink that prevent HTTP Keep-Alive working too well, the LWP::UserAgent instances aren't always reused. That may cause the delay even between requests to the same hostname during a single checklink run.
Received on Thursday, 9 October 2003 14:24:55 UTC