lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dawid Weiss <>
Subject [IMPORTANT] Thread leak patch
Date Sat, 11 Aug 2012 10:31:44 GMT

This is important to avoid frustrations, please read until the end.
I've been working in the background for some time now and I think I am
ready to commit the long-planned runner changes. It includes:

- timeout control (entire suite must not last longer than 10 minutes),
- thread leak control (threads must not leak outside of a test suite;
there is an ignore filter with certain system threads and static
thread pool threads),
- completely refactored output logging; -Dtestcase=... will show the
test output immediately (passthrough mode), concurrent runs should not
OOM as well as long as you have enough disk space for a spill buffer.

I've been running two build servers on the trunk for some time now and
it's incredibly (suspiciously) stable with the latest changes -- about
30 runs on windows and linux without any failure. Here's the caveat: I
did exclude a few tests that I couldn't fix and that were known to
cause problems. So here's my proposal:

1) I will commit this patch to the trunk WITH @Ignore ANNOTATION on
those suites that currently don't pass the tests (because they leak
zombie threads for example). These are also marked @BadApple so
they're easy to find using Eclipse/ IntelliJ. I don't have enough
knowledge to fix these tests but I've spoken to Mark and I'm willing
to provide any help I can to make these tests pass. Until they do
pass, however, I would exclude them from the build.

2) This is a major refactoring of the runner and it took me some time
to stabilize it. It's been tested extensively but I cannot exclude the
possibility that some VM/OS combinations will act weird. Please test
and in case you see something odd, report back to me. Include *.events
files and any logs if possible.

3) Once we have the trunk running I would port this patch to 4.x
branch. Alternatively, we can do it right away -- I am pretty
confident it's for the better.

I look forward to hearing your opinion on the above. If I hear no
vetos I will apply this patch tomorrow in the afternoon (Sunday) which
will give me some time to revert in case I see trouble. I chose the
weekend because I believe most of you will read this e-mail and will
not care much (it's a weekend after all).


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message