Re: [Boost-testing] [boost] [regression] Tests for 'develop' not showing

Subject: Re: [Boost-testing] [boost] [regression] Tests for 'develop' not showing
From: Tom Kent (lists_at_[hidden])
Date: 2015-09-23 18:52:06

On Wed, Sep 23, 2015 at 8:16 AM, Tom Kent <lists_at_[hidden]> wrote:

>
> On Tue, Sep 22, 2015 at 9:53 PM, Rene Rivera <grafikrobot_at_[hidden]>
> wrote:
>>
>> > On Wed, Sep 9, 2015 at 4:35 PM Edward Diener <eldiener_at_[hidden]>
>>>> > wrote:
>>>> >
>>>> > > Trying
>>>> > >
>>>> http://www.boost.org/development/tests/develop/developer/summary.html I
>>>> > > get:
>>>> > >
>>>> > > "Error extracting file: The specified zipfile was not found."
>>>> > >
>>>>
>>>> And they are messed up again. Tom can you look into your side of this?
>> When I generated the results I got the ZIP files to be 1.5GiB and 2.0GiB,
>> approximately, for develop and master, respectively. Yet what I see on the
>> web site now is 1.5GiB and 200MiB. Which is clearly incorrect. And it's
>> precisely the shorter ZIP, i.e. master.zip, that now doesn't work. I.e
>> browsing develop results works, but not browsing master results.
>>
>
> Over the last week I've seen both the master and develop results showing
> the zip error message.
>
> I just noticed that in the last master run, I had a segfault in the
> processing:
>
> tomkent_at_eddie:~/tmp/boost_reports/boost-reports⟫ tail master.log
> Reading failures markup
> Reading
> /home/tomkent/tmp/boost_reports/boost-reports/master/incoming/master/processed/BP
> x86_64 C++11.xml
> Merging expected results
> Generating links pages
> Reading
> /home/tomkent/tmp/boost_reports/boost-reports/master/incoming/master/processed/CrystaX.NET-apilevel-19-armeabi-v7a-hard.xml
> Merging expected results
> Generating links pages
> Reading
> /home/tomkent/tmp/boost_reports/boost-reports/master/incoming/master/processed/CrystaX.NET-apilevel-19-armeabi-v7a.xml
> Merging expected results
> Segmentation fault (core dumped)
>
> I'll look into it more tonight.
> Tom
>

It looks like running the reports (esp master) is taking much longer...up
to 2hrs. I had a cron job running every hour, which corrupted things when
it was kicked off a second time while the first was still running.

I've made a change in the script to setup a lock file, so that only one
instance will run and submitted a pull request against the regression repo.
It's already running on my report generator, so hopefully the problem won't
show up again. If anyone sees it please let me know.

Tom