AW: AW: AW: bzr selftest (on solaris 10): too may open files

[Vincent Ladeuil]

>>>>> "Bernhard" == Voelker, Bernhard <bernhard.voelker at siemens.com> writes:

    Bernhard> Hello Vincent,
    Bernhard> Vincent wrote:

    >> >>>>> "Bernhard" == Voelker, Bernhard <bernhard.voelker at siemens.com>
    Bernhard> writes:
    >> 
    Bernhard> Is python 2.4.4 supposed to work with bzr?
    >> 
    >> Yes. But the stable 2.4 version is 2.4.5, so may be you suffer
    >> from a python bug. Is upgrading to 2.4.5 or even better to 2.5
    >> an option in your case ?

    Bernhard> Well, I could ask my root, but from reading the few changes in 2.4.5
    Bernhard> I don't expect much better results:
    Bernhard> http://python.org/download/releases/2.4.5/NEWS.txt 

Hmmm, indeed.

    >> ...

    >> My suggestion would then be to try to run smaller parts of the
    >> test suite to identify the first failing test, or just run the
    >> test suite with the '-1' flag:
    >> 
    >> bzr selftest -1

    Bernhard> Logfile "bzr-selftest-1.log" attached.
    Bernhard> Unfortunately a coredump happened ;-(

Weird. Did you try to build the C extensions ? How did you install bzr ?

    Bernhard> Need more?

See below.

    Bernhard> BTW: Using bzr seems to work - for my very little, local repositories
    Bernhard> until now.


You can try 'bzr selftest -s bb' which will run the black box
tests, they are significant subset of the test suite and should
gives you a better feeling of what parts of bzr are expected to
work.

I expect that most (if not all) of them should pass. 


    Bernhard> Bye,
    Bernhard> Berny


    Bernhard> tests failed
    Bernhard> bzrlib.tests.blackbox.test_branch.TestBranchStacked.test_branch_stacked_from_smart_server is leaking threads among 5 leaking tests
    Bernhard> testing: /vol1/home1/vb027591/bzr/bzr-1.9/bzr
    Bernhard>    /vol1/home1/vb027591/bzr/bzr-1.9/bzrlib (1.9 python2.4.4)

<snip/>
    Bernhard> ^^^^[log from bzrlib.tests.blackbox.test_diff.TestExternalDiff.test_external_diff]
    Bernhard> ----------------------------------------------------------------------
    Bernhard> Traceback (most recent call last):
    Bernhard>   File "/vol1/home1/vb027591/bzr/bzr-1.9/bzrlib/tests/blackbox/test_diff.py", line 362, in test_external_diff
    Bernhard>     self.assertEqual('', err)
    Bernhard> AssertionError: not equal:
    Bernhard> a = ''
    Bernhard> b = "diff: illegal option -- label\nusage: diff [-bitw] [-c | -e | -f | -h | -n | -u] file1 file2\n       diff [-bitw] [-C number | -U number] file1 file2\n       diff [-bitw] [-D string] file1 file2\n       diff [-bitw] [-c | -e | -f | -h | -n | -u] [-l] [-r] [-s] [-S name] directory1 directory2\nbzr: ERROR: external diff failed with exit code 2; command: ['diff', '--label', u'goodbye\\t1970-01-01 00:00:00 +0000', '/tmp/bzr-diff-old-OdpSol', '--label', u'goodbye\\t2008-11-18 11:54:03 +0000', '/tmp/bzr-diff-new-EyDjsJ', '--binary', '-ub']\n"

Well, that one expects a GNU diff I think (reasonably recent) so
it's not really relevant :-/

I don't have access to a Solaris machine anymore, but IIRC bzr
was running reliably there. It was on a Sparc with a preview
release of Solaris 10. It was also working reliably on Solaris 8.

But in both cases I'm pretty sure I had a GNU diff somewhere in
my path to avoid the above problem.

I'd like to have the full test suite running for Solaris but the
best I can do for now is to setup a virtual machine with open
solaris and I'm not sure it will be that relevant eort produce
your problems.

Can you tell us what hardware/software you're running ?

    Vincent