Re: [Boost-testing] HEAD testing wedged?

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2006-11-03 15:39:45

Martin Wille writes:
> Eric Niebler wrote:
> > I know everyone is focused 110% on 1.34, but I don't suppose someone can 
> > kick the machine running the tests against HEAD? When I go to 
> > http://engineering.meta-comm.com/boost-regression/CVS-HEAD/developer/index.html, 
> > I see:
> > 
> > Warning: CVS snapshot is 115 hours old.
> 
> There's no 'the' machine running tests. You apparently are talking
> about the machine that combines the result sets from all testing
> machines.

And that warning doesn't mean that the _report generation_ is
stalled, just that the hourly CVS tarball we provide for
'regression.py' has not been timely updated.

> 
> I guess this is a communication (firewall? ssh key?) problem between 
> meta-comm and SF. The 1.34 page contains a similar message.

Yes, our machine ran out of space --
http://article.gmane.org/gmane.comp.lib.boost.testing/3470.

> Perhaps, we should have second machine performing the same service.

It wouldn't hurt.

> Are there any instructions about how to build the result tables?

1. Create a working directory, say '~/build_reports'.

2. Inside the working directory, checkout a local copy of Boost
   sources from the CVS (HEAD would do) so that, say, 
   '~/build_reports/boost/libs/expected_results.xml' is a valid path.

3. Download
   http://engineering.meta-comm.com/boost/build_reports.tar.bz2 and
   place its content inside the working directory.

4. To start one cycle of report generation, do:

    /usr/bin/python ~/build_reports/script_runner.py \
      --failure-mail=<your email> ~/build_reports/build_reports.py \
      --user=<your SF user name> --source=CVS --cvs-tag=HEAD \
      --website-dir=<destination dir to copy the report pages archive to>

5. Pester us if something doesn't work :).

HTH,
-- 
Aleksey Gurtovoy
MetaCommunications Engineering