[Python-Dev] Strange umask(?)/st_mode issue

[Cameron Simpson]

On 29Mar2019 19:30, Steve Dower <steve.dower at python.org> wrote:
>On 29Mar.2019 1731, Nathaniel Smith wrote:
>> That does sound strange.
>
>Oh good, it's not just me :)
>
>> How easily can you reproduce it? That majorly
>> effects how I would try to debug something like this...
>
>No idea. Looking at the builds on Pipelines again, it seems that all the
>builds since about 10am PDT have been failing, so perhaps they rolled
>out something that changed how umask works? I don't have any idea
>whether that's possible though.
>
>> If you're able to get an strace of a failed run then that would probably
>> tell us a *lot*.
>
>I'll need help with that :) But if it's just commands in the
>.azure-pipelines/posix-steps.yml file then sending a PR with them added
>should do it.

Can you get a branch into your pipeline? Then you could just hack the 
tarfile test with something quick and dirty like:

    pid = os.getpid()
    system("strace -p %d 2>/path/to/strace.out &" % pid)
    time.sleep(2)   # get strace heaps of time to start

just before the tarfile open. A ghastly hack but it would get you 
debugging info. You could even decide to remove the strace.out file if 
the umask issue doesn't show, if it is erratic (can't see why it would 
be though).

Cheers,
Cameron Simpson <cs at cskk.id.au>