| Summary: | gef-nightly-tycho fails because of ClassLoader problem within Hudson | ||
|---|---|---|---|
| Product: | Community | Reporter: | Alexander Nyßen <nyssen> |
| Component: | CI-Jenkins | Assignee: | Eclipse Webmaster <webmaster> |
| Status: | RESOLVED NOT_ECLIPSE | QA Contact: | |
| Severity: | blocker | ||
| Priority: | P3 | ||
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | All | ||
| Whiteboard: | |||
|
Description
Alexander Nyßen
Slave1 has been restarted. It would appear that increasing the number of executors (on a given slave) causes this problem to occur more often. I'm going to close as 'not_eclipse' since this actually seems to be a Hudson issue. -M. (In reply to comment #1) > Slave1 has been restarted. It would appear that increasing the number of > executors (on a given slave) causes this problem to occur more often. > > I'm going to close as 'not_eclipse' since this actually seems to be a Hudson > issue. > > -M. Agreed, but thats the second or third time the gef build ran into this problems. Do you have any idea what is the underlying cause. Is there a change we can find out and report it? I don't have any special insight(not being a Java wizard), but I do recall seeing others(via Google) with this issue. My simplistic guess is that the 'widget' that tracks slave activity/execution is easily corrupted when the number of executor threads on a slave starts to climb(and one of the threads falls over). We're working with the Hudson folks to get them some debug access so hopefully they may be able to explain this(or even fix it) -M. |