test_bufio fails sporadically on the AMD64 Windows7 SP1 buildbot, e.g.: http://buildbot.python.org/all/builders/AMD64%20Windows7%20SP1%203.x/builds/161/steps/test/logs/stdio test_nullpat (test.test_bufio.CBufferSizeTest) ... ok test_primepat (test.test_bufio.CBufferSizeTest) ... ERROR test_nullpat (test.test_bufio.PyBufferSizeTest) ... ok test_primepat (test.test_bufio.PyBufferSizeTest) ... ok ====================================================================== ERROR: test_primepat (test.test_bufio.CBufferSizeTest) ---------------------------------------------------------------------- Traceback (most recent call last): File "C:\Users\Buildbot\buildbot.python.org\3.x.kloth-win64\build\lib\test\test_bufio.py", line 59, in test_primepat self.drive_one(b"1234567890\00\01\02\03\04\05\06") File "C:\Users\Buildbot\buildbot.python.org\3.x.kloth-win64\build\lib\test\test_bufio.py", line 52, in drive_one self.try_one(teststring) File "C:\Users\Buildbot\buildbot.python.org\3.x.kloth-win64\build\lib\test\test_bufio.py", line 20, in try_one support.unlink(support.TESTFN) File "C:\Users\Buildbot\buildbot.python.org\3.x.kloth-win64\build\lib\test\support.py", line 229, in unlink os.unlink(filename) PermissionError: [Error 5] Access is denied: '@test_2420_tmp' Is there some kind of anti-virus installed on the machine?
> Looks similar to the sporadic test_reprlib failure. Roumen has explained > that here: I don't think there's anything similar. The path is much smaller in test_bufio.
I've changed the malware scanner to disable real-time protection and have disabled scanning of the buildbot directories. Note that I could reproduce intermittent failures when using '-j' of regrtest, but with these changes they no longer appear.
> I've changed the malware scanner to disable real-time protection and > have disabled scanning of the buildbot directories. Thank you. I hope that, thanks to this change, we later can make the Win64 buildbot part of the bunch of stable buildbots.
The implementation of test.support.unlink() was changed since. It tries to change permissions of read-only files, and repeat attempts several times. It looks more reliable now. Is this issue still reproducible?