cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Philip Thompson (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-12957) Segfault shadows failures in unit tests
Date Mon, 28 Nov 2016 17:26:59 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-12957?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15702564#comment-15702564
] 

Philip Thompson commented on CASSANDRA-12957:
---------------------------------------------

Is there ever an instance in which a segfault occurs but we don't see 
{{Forked Java VM exited abnormally. Please note the time in the report does not reflect the
time until the VM exit.}}?

I think we need to just be more vigilant about examining those. We're limited to consuming
what junit outputs, and we should start always looking for segfaults in the logs when we see
these types of errors.

> Segfault shadows failures in unit tests
> ---------------------------------------
>
>                 Key: CASSANDRA-12957
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12957
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Testing
>            Reporter: Alex Petrov
>
> It looks like a segfaulting VM will shadow the CI results, for example [here|https://cassci.datastax.com/view/Dev/view/ifesdjeen/job/ifesdjeen-12651-segfault-testall/1/],
I added a test that segfaults (on 
> Some times the tests with
> {code}
> Forked Java VM exited abnormally. Please note the time in the report does not reflect
the time until the VM exit.
> {code}
> error messages are disregarded or skipped due to the lack of information. Although if
you look closer in the [full log|https://cassci.datastax.com/view/Dev/view/ifesdjeen/job/ifesdjeen-12651-segfault-testall/1/consoleFull]
you will see the actual segfault: 
> {code}
> 12:51:30     [junit] # A fatal error has been detected by the Java Runtime Environment:
> 12:51:30     [junit] #
> 12:51:30     [junit] #  SIGSEGV (0xb) at pc=0x00007f7dac4b9900, pid=12587, tid=140177756509952
> {code}
> Test result is {{1 failure}} (which is kind of expected), but this failure resulted into
skip of all the other tests. 
> Can we track the tests that were skipped not by test runner (reported by {{Failed: x,
Skipped: y}} in test results), but without "intention" (test runner crash, segfault etc) and
maybe add more infos on segfaults (possibly whole trace)? Might be there are unnoticed segfaults?
> cc [~philipthompson]



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message