!buildbot windows

🤖 New build scheduled with the buildbot fleet by @zooba for commit 8b29604 🤖

The command will test the builders whose names match following regular expression: windows

The builders matched are:

• AMD64 Windows11 Refleaks PR
• ARM64 Windows Non-Debug PR
• ARM64 Windows PR
• AMD64 Windows Server 2022 NoGIL PR
• AMD64 Windows11 Non-Debug PR
• AMD64 Windows11 Bigmem PR
• AMD64 Windows10 PR

Buildbots look fine, the one failure is unrelated and doesn't repro across the rest, so I'm choosing to ignore it.

Thanks @zooba for the PR 🌮🎉.. I'm working now to backport this PR to: 3.12.
🐍🍒⛏🤖

GH-113989 is a backport of this pull request to the 3.12 branch.

Not sure this is the best place to add this, but since this change it appears that my Windows 10 buildbot consistently fails the new test during the main run, but passes upon retry. I'm not sure if it indicates an issue with the test or the environment, but it's never passed at any point the first time, which seems odd - maybe some sort of race that only shows up when the tests are being run in parallel?

The same thing started happening on the 3.12 branch as soon as it was backported.

@db3l Can you link one of these failures? Or copy in the error text

Sorry, my bad. Here's a recent 3.x run:
https://buildbot.python.org/all/#/builders/146/builds/7436

and the actual failure during the main run (which seems a true failure, as without access still gets a result):

======================================================================
FAIL: test_stat_inaccessible_file (test.test_os.Win32NtTests.test_stat_inaccessible_file)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "D:\buildarea\3.x.bolen-windows10\build\Lib\test\test_os.py", line 3134, in test_stat_inaccessible_file
    self.assertEqual(0, stat2.st_dev)
    ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
AssertionError: 0 != 17153176533692772311
Stdout:
File: @test_6260_tmp�
stat with access: os.stat_result(st_mode=33206, st_ino=844424930132043114, st_dev=17153176533692772311, st_nlink=1, st_uid=0, st_gid=0, st_size=9, st_atime=1705336060, st_mtime=1705336060, st_ctime=1705336060)
 without access: os.stat_result(st_mode=33206, st_ino=844424930132043114, st_dev=17153176533692772311, st_nlink=1, st_uid=0, st_gid=0, st_size=9, st_atime=1705336060, st_mtime=1705336060, st_ctime=1705336060)

The retry at the end of the test run always seems to succeed, but even though the runner says it's retrying in verbose mode I don't get the output from the test, but I expect that could just be an issue in the retry logic.

0:20:55 load avg: 0.22 Re-running 1 failed tests in verbose mode in subprocesses
0:20:55 load avg: 0.22 Run 1 test in parallel using 1 worker process (timeout: 20 min, worker timeout: 25 min)
0:20:56 load avg: 0.22 [1/1] test_os passed
Re-running test_os in verbose mode (matching: test_stat_inaccessible_file)
test_stat_inaccessible_file (test.test_os.Win32NtTests.test_stat_inaccessible_file) ... ok
----------------------------------------------------------------------
Ran 1 test in 0.059s
OK

I've also since noticed that Itamar's windows buildbot seems to be getting the same results, so a test issue seems more likely than buildbot - perhaps something with the icacls (actual execution, or latency) during the regular portion of the run that isn't a subprocess failure? The test itself seems fine to me at first glance otherwise.

Yeah, looks like it's failing to reset permissions properly. I wonder if it's because of the Unicode filename, or possibly because it's not using a fully qualified path? I'll try making a full path first, I guess.

I'll ping you on a new PR and we can continue there.