[Bug general/4731] New: Junit going crazy after test failure

[Bug general/4731] New: Junit going crazy after test failure

[kris dot van dot hees at oracle dot com]

An error being reported as a result of the testLeakyPipes() test in
TestFileDescriptor seems to have thrown junit into a fit, causing rather weird
behaviour (such as what seems like concurrent streams writing output causing
very intermixed output, with erratic reporting of what should have been final
results of the entire suite being doing on an almost by-file basis).

See the log at the following URL:
http://build.alchar.org/~aedil/rep/ca-tools2/dl/frysk_incr.20070702-014501.log.bz2

-- 
           Summary: Junit going crazy after test failure
           Product: frysk
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: general
        AssignedTo: frysk-bugzilla at sourceware dot org
        ReportedBy: kris dot van dot hees at oracle dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=4731

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug general/4731] Junit going crazy after test failure

[cagney at redhat dot com]

------- Additional Comments From cagney at redhat dot com  2007-07-03 17:03 -------
Do you know what caused the error in testLeakyPipes (I can't repo this on FC-5)?


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4731

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug general/4731] Junit going crazy after test failure

[kris dot van dot hees at oracle dot com]

------- Additional Comments From kris dot van dot hees at oracle dot com  2007-07-03 17:58 -------
Well, if you look at the code, you see that the exception (Too many open files)
is thrown from the Pipe object instantiation (in the native code) when
tryGarbageCollect() is called repeatedly within a single attempt to create a
pipe.  So, clearly there was a problem with the garbage collection process not
being able to ensure that there were fds available for creating the pipe.

Regardless of the reason why this happened, junit shouldn't blow up over it.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4731

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug general/4731] Junit going crazy after test failure

[cagney at redhat dot com]

------- Additional Comments From cagney at redhat dot com  2007-07-03 19:45 -------
(In reply to comment #2)
> Well, if you look at the code, you see that the exception (Too many open files)
> is thrown from the Pipe object instantiation (in the native code) when
> tryGarbageCollect() is called repeatedly within a single attempt to create a
> pipe.  So, clearly there was a problem with the garbage collection process not
> being able to ensure that there were fds available for creating the pipe.

Right.  Is there any external factor that could be triggering this difference in
behavour?

> Regardless of the reason why this happened, junit shouldn't blow up over it.

While one is disputing that JUnit and the JVM should handle this more
gracefully, running out of FDs is an extreem situtation and not well-defined so
we may need to cut some slack here.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=4731

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.