On Thu, Mar 20, 2008 at 5:22 AM, Senthil Kumaran S <senthil_at_collab.net> wrote:
> Lieven Govaerts wrote:
> > Relevant extract from apache error log:
> > [Thu Mar 20 09:56:19 2008] [error] [client 127.0.0.1] The URI does not
> > contain the name of a repository. [403, #190001]
> >
> > access.log:
> > 127.0.0.1 - jrandom [20/Mar/2008:09:56:19 +0100] "OPTIONS
> > /svn-test-work/repositories HTTP/1.1" 403 322 "-" "SVN/1.6.0 (dev build)
> > neon/0.25.5"
> >
> > So, why is there no repository name?
>
> I could see in all the test failures stacktrace we have this line:
>
> <snip>
>
> File
> "/home/stylesen/subversion-dev/trunk/subversion/tests/cmdline/svntest/main.py",
> line 436, in run_command_stdin
> raise Failure
> </snip>
>
> which in turn gives a 403 response in apache. The doc string of
> run_command_stdin says:
>
> " this should not be very large, as if the program outputs more than the OS
> is willing to buffer, this will deadlock, with both Python and
> COMMAND waiting to write to each other for ever"
>
> So my assumption is when the tests are run in parallel the STDIN_LINES created
> by different processes are large and raises a failure randomly for the above
> reason?
I'm getting this issue as well.
I'm not sure if I follow your diagnosis. The different parallel tests
are in separate processes (the only threading is in a dispatching
process); I don't see why this would make the STDIN_LINES larger than
one process. Plus we're clearly getting a 403 from the server.
In the past, this has been because we've mixed in authz-using tests
(which can't be run in parallel) with other tests. I don't see that
here though.
--dave
--
David Glasser | glasser@davidglasser.net | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-03-31 23:06:36 CEST