Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 402846

Summary: slow or no access to p2 repositories on download.eclipse.org from Hudson ?
Product: Community Reporter: Matthias Sohn <matthias.sohn>
Component: CI-JenkinsAssignee: Eclipse Webmaster <webmaster>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: P3 CC: david_williams, john.arthorne, ken.lee, robin, tomasz.zarna, webmaster
Version: unspecified   
Target Milestone: ---   
Hardware: PC   
OS: Mac OS X   
Whiteboard:
Bug Depends on: 401964    
Bug Blocks:    

Description Matthias Sohn CLA 2013-03-10 18:28:17 EDT
It seems like Hudson can't access p2 repositories on download.eclipse.org anymore (at least on the slaves I tried) maybe some wrong proxy settings in my way or is it just very slow ?
The egit job fails since Mar 2 often due to build timeouts

https://hudson.eclipse.org/hudson/job/egit/2790/consoleFull
https://hudson.eclipse.org/hudson/job/egit/2788/consoleFull

formerly the job used to finish within 20-25 min, even increasing max build time to 60 min didn't help.
Comment 1 Matthias Sohn CLA 2013-03-10 18:30:52 EDT
See build time trend of the egit job
https://hudson.eclipse.org/hudson/job/egit/buildTimeTrend
Comment 2 David Williams CLA 2013-03-11 03:39:04 EDT
My theory is all requests are going though the proxies ... see bug 401964. 
(And, pretty tough to get p2 to work through a proxy).
Comment 3 Matthias Sohn CLA 2013-03-11 18:46:00 EDT
I tried again with 
-Djava.net.useSystemProxies=true
and this still fails when trying to download Jetty 7.6 from the archive server:

!MESSAGE Connection to http://archive.eclipse.org/jetty/updates/jetty-bundles-7.x/7.6.0.v20120127/p2.index failed on Connection timed out. Retry attempt 0 started

see https://hudson.eclipse.org/hudson/job/egit/2795/console
Comment 4 Matthias Sohn CLA 2013-03-11 20:40:21 EDT
the egit job doesn't hit this network problem on the master node, but on none of the following slaves I tried: slave1, slave2, slave4, slave6, fastlane

So it seems something is wrong with the network settings of the slaves.

Unfortunately on master some tests can't cleanup resources created during the tests, also wiping the Hudson workspace fails with an error message. Since wiping the workspace on master failed partially now git refuses to fetch so I am now completely stuck with the egit job.
Comment 5 Eclipse Webmaster CLA 2013-03-12 15:42:34 EDT
I updated the internal DNS and tweaked the proxy rules for slave1 and it seems to be happy to run the egit job.  If you can verify this I'll roll these changes to the other slaves.

-M.
Comment 6 Matthias Sohn CLA 2013-03-12 19:53:56 EDT
(In reply to comment #5)
> I updated the internal DNS and tweaked the proxy rules for slave1 and it
> seems to be happy to run the egit job.  If you can verify this I'll roll
> these changes to the other slaves.

looks good, please copy these settings to all other slaves
Comment 7 Eclipse Webmaster CLA 2013-03-13 11:00:49 EDT
Ok, I"ve pushed these changes to the other slaves.

-M.
Comment 8 Denis Roy CLA 2013-03-15 10:35:55 EDT
grrr, changed the wrong bug.