On 24 November 2010 08:08, Niklas Gustavsson <nik...@protocol7.com> wrote: > On Tue, Nov 23, 2010 at 1:15 PM, Niklas Gustavsson <nik...@protocol7.com> > wrote: >> On Tue, Nov 23, 2010 at 12:03 PM, sebb <seb...@gmail.com> wrote: >>> https://hudson.apache.org/hudson/job/JMeter-trunk/1182/console >>> >>> Build Total time: 3 minutes 40 seconds >>> Overall Took 30 min on solaris1 >> >> Yes, this has been reported before but I was unable to find the cause >> at that time. Any ideas to explore are welcome :-) > > I've been looking at the network traffic on the Hudson master, and > we're nowhere near being saturated on that box. On the other hand, > slow archiving seems to be a well known problem with Hudson: > http://issues.hudson-ci.org/browse/HUDSON-3922 > http://issues.hudson-ci.org/browse/HUDSON-7813 > http://issues.hudson-ci.org/browse/HUDSON-3799 > http://issues.hudson-ci.org/browse/HUDSON-7921 > > Also, the number of builds running on Hudson has increased with some > 25% in the last two months or so, meaning things are getting a bit > crowded. I'm suspecting these might be adding up to the problems we're > currently seeing. > > I've added a comment on 3922 asking if there is any ongoing work in > fixing this. I'll let you know if I get any answer.
I just discovered that I was accidentally using http://hudson.zones.apache.org/ instead of https://hudson.apache.org/ to download some JMeter build files. Instead of taking about 10 seconds, the downloads took several minutes. Is it possible that the wrong URL is being used in Hudson somewhere? I'm sure the http: URL used to work OK, but I'm not sure how long ago that was - possibly a month or more. > /niklas >
