this with the patch applied. I think this
should be committed in as soon as possible?
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project
[
https://issues.apache.org/jira/browse/LUCENE-3855?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael McCandless resolved LUCENE-3855.
Resolution: Fixed
TestStressNRT failures (reproducible
...
Basically I added a refCount to ReadersAndLiveDocs, and
incRef/decRef'd in the right places. I also pulled out
ReadersAndLiveDocs to its own source, and tried to simplify
it a bit (separate method calls to dropChanges,
drop, release).
TestStressNRT failures (reproducible
to its own source
{quote}
+1
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
we need to add a
separate refCount to RLD to fix this... I'm working on that.
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
this looks like a race condition between closing the writer and a concurrent
merge scheduler?
Let me know if you need any further stacks/ tracing listings -- this is fairly
easy to reproduce on my machine and I can add anything.
TestStressNRT failures (reproducible
:)
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
Reporter: Dawid Weiss
)
{noformat}
Uwe, this just calls for your German analytic mind to solve :)
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene
the failure too, if I beast the test...
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
in the background doing just while
(true);. I have a slower computer 2-core computer and this happens there most
often.
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira
Not a linux box, but -Dtests.iter=1000 and no failures on my
Macbook Pro.
FWIW
Erick
On Wed, Mar 7, 2012 at 5:52 PM, Dawid Weiss dawid.we...@gmail.com wrote:
https://issues.apache.org/jira/browse/LUCENE-3855
Ehm, either I'm going crazy or my machines are somehow different from
anybody else's
I also don't get any failures. I ran the test on a linux box (latest
checkout) and with -Dtests.iter=1000
Martijn
On 8 March 2012 13:45, Erick Erickson erickerick...@gmail.com wrote:
Not a linux box, but -Dtests.iter=1000 and no failures on my
Macbook Pro.
FWIW
Erick
On Wed, Mar 7, 2012
@lucene.apache.org
Subject: Re: TestStressNRT failures.
Not a linux box, but -Dtests.iter=1000 and no failures on my Macbook Pro.
FWIW
Erick
On Wed, Mar 7, 2012 at 5:52 PM, Dawid Weiss dawid.we...@gmail.com wrote:
https://issues.apache.org/jira/browse/LUCENE-3855
Ehm, either I'm going crazy or my machines
I also don't get any failures. I ran the test on a linux box (latest
checkout) and with -Dtests.iter=1000
Did you run with any of the known failing seeds though? For example this one?
-Dtests.seed=69468941c1bbf693:19e66d58475da929:69e9d2f81769b6d0
-Dargs=-Dfile.encoding=UTF-8
Dawid
I let it run for 100 or so iterations in a shell while loop, and I got zero
failures.
I couldn't repeat it on a similar Windows box as well.
Dawid
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional
[
https://issues.apache.org/jira/browse/LUCENE-3855?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael McCandless reassigned LUCENE-3855:
--
Assignee: Michael McCandless
TestStressNRT failures (reproducible
method, and when I run that indeed I see the unhandled
exception from thread. So I'm not sure why exceptions in CMS threads in
particular are not being reported...
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
.
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
Reporter: Dawid Weiss
Assignee
.
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
Reporter: Dawid Weiss
,
because IW closes CMS first (which should join to all outstanding threads),
before hitting the exc in main thread. Still baffled...
Anyway it's not a blocker for me... I'm printing the exc in CMS.handleExc.
TestStressNRT failures (reproducible
something about it.
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
I got a failure on second run:
[junit] - Standard Error -
[junit] NOTE: reproduce with: ant test -Dtestcase=TestStressNRT
-Dtestmethod=test
-Dtests.seed=69468941c1bbf693:19e66d58475da929:69e9d2f81769b6d0
-Dargs=-Dfile.encoding=UTF-8
[junit] NOTE: test
The test seems to fail in different ways, here's couple more examples:
[junit] - Standard Error -
[junit] NOTE: reproduce with: ant test -Dtestcase=TestStressNRT
-Dtestmethod=test
-Dtests.seed=69468941c1bbf693:19e66d58475da929:69e9d2f81769b6d0
this line:
/home/hossman/lucene/dev/lucene/common-build.xml:677: Tests failed!
Total time: 8 seconds
TOTAL LOOPS: 52
{noformat}
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https
The test seems to fail in different ways, here's couple more examples:
Yes, it does. They all seem to be related to CMS though -- I attached
a few stacks to the issue.
Thanks for helping Sami (and everyone).
Dawid
-
To
always hold wrt
multi-threaded access? I mean something that I could try to weave into the code
so that we can catch an abnormal interaction pattern while it's happening (as
opposed just seing the result)?
TestStressNRT failures (reproducible
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
Reporter: Dawid Weiss
?
Doesn't reproduce on my machine... does it need multiple iterations usually to
fail?
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
quite frequently (I re-run the
test from scratch, didn't use repetitions). Interestingly, the problem seems to
be different every time. I'll attach logs.
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
output1.log
Four executions, four errors (same seed).
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
}
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
Reporter: Dawid Weiss
Priority
] at
org.apache.lucene.util.SystemPropertiesInvariantRule$1.evaluate(SystemPropertiesInvariantRule.java:22)
[junit]
[junit]
[junit] Test org.apache.lucene.index.TestStressNRT FAILED
{noformat}
TestStressNRT failures (reproducible
] at
org.apache.lucene.util.SystemPropertiesInvariantRule$1.evaluate(SystemPropertiesInvariantRule.java:22)
[junit]
[junit]
[junit] Test org.apache.lucene.index.TestStressNRT FAILED
{noformat}
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
https://issues.apache.org/jira/browse/LUCENE-3855
Ehm, either I'm going crazy or my machines are somehow different from
anybody else's ;)
If you have a spare cycle could you:
cd lucene
ant test-core -Dtestcase=TestStressNRT -Dtestmethod=test
}
TestStressNRT failures (reproducible)
-
Key: LUCENE-3855
URL: https://issues.apache.org/jira/browse/LUCENE-3855
Project: Lucene - Java
Issue Type: Bug
Reporter: Dawid Weiss
35 matches
Mail list logo
