Re: [PATCH v4 3/3] refs/reftable: reload locked stack when preparing transaction

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 1, 2024, 07:34 UTC
Message-ID: <xmqqjzes2s5y.fsf@gitster.g>
In-Reply-To: <c4lz3begoplgde5iimvk4k7cufiyryntccqo46u3fy5qvqauv3@tta5wfg2ik5t>

Josh Steadmon <steadmon@google.com> writes:

Show 25 quoted lines

> On 2024.09.27 00:07, Jeff King wrote:
>> On Tue, Sep 24, 2024 at 07:33:08AM +0200, Patrick Steinhardt wrote:
>> 
>> > +test_expect_success 'ref transaction: many concurrent writers' '
>> > +	test_when_finished "rm -rf repo" &&
>> > +	git init repo &&
>> > +	(
>> > +		cd repo &&
>> > +		# Set a high timeout such that a busy CI machine will not abort
>> > +		# early. 10 seconds should hopefully be ample of time to make
>> > +		# this non-flaky.
>> > +		git config set reftable.lockTimeout 10000 &&
>> ...
>
> We're seeing repeated failures from this test case with ASan enabled.
> Unfortunately, we've only been able to reproduce this on our
> $DAYJOB-specific build system. I haven't been able to get it to fail
> using just the upstream Makefile so far. I'll keep trying to find a way
> to reproduce this.
>
> FWIW, we're not getting I/O errors, we see the following:
> fatal: update_ref failed for ref 'refs/heads/branch-20': cannot lock references
>
> We tried increasing the timeout in the test to 2 minutes (up from 10s),
> but it didn't fix the failures.

Thanks for a report, and please keep digging ;-).

Is your build, like Peff's, for Windows, or your variant of Linux?

Thanks.