[
https://issues.apache.org/jira/browse/IVY-598?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12523147
]
Xavier Hanin commented on IVY-598:
----------------------------------
This is really strange. According to the dumps, it seems the problem occurs
during Ant classloading mechanism. The debug output seems to confirm that (no
"Class org.apache.ivy.core.retrieve.RetrieveEngine loaded from ant loader
(parentFirst)" output at the end). Moreover all thread dumps seems to be stuck
at the same line of Ivy class: 170. This is the line where Ivy references the
RetrieveEngine. So this classloading problem seems to happen always with the
RetrieveEngine class.
Hence here is what I would try:
- use another version of Ant (1.6.5 for instance) to double check this is not a
problem related to the Ant version you're using
- download Ivy from Apache again, verify the checksum and signature, and put it
somewhere else on your disk, and update your build to use this freshly
downloaded version. You may have a problem of file corruption in the jar you're
using, or a problem with your hard disk.
- the problem does not seem to be related to IvyInfo itself, but to Ivy
initialization. I would try to confirm that by loading the settings explicitly
as first operation (before ivy:info)
- you seem to load ivy.jar using a classpath defined in your build. I would
also try other mechanism, like putting ivy jar in your ant lib directory
Depending on the results of these tests, we may get a better understanding of
the problem...
> ivy:info locks up
> -----------------
>
> Key: IVY-598
> URL: https://issues.apache.org/jira/browse/IVY-598
> Project: Ivy
> Issue Type: Bug
> Components: Ant
> Affects Versions: 2.0.0-alpha-2
> Environment: OSX 10.4.10, JDK 1.5.0_07
> Reporter: Colin Fleming
> Attachments: debug.output.txt, threaddump.txt, threaddump2.txt,
> threaddump3.txt
>
>
> Occasionally Ivy seems to deadlock, and something that I've changed in our
> project suddenly makes it happen all the time. This is an urgent issue for
> us, it used to only affect me from time to time but suddenly it's frequently
> happening to most people in our project. It appears to lock up loading
> classes during an ivy:info. I can't break out with a Ctrl-C, I have to kill
> -9 the process from another terminal. I've attached three thread dumps from
> different times it's happened plus the debug output of one run of the
> ivy:info task (the run associated with threaddump3, in case it helps). This
> is almost 100% reproducible for me now, which unfortunately means I can build
> almost 0% of the time :(
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
