Re: t5559 breaks with apache 2.4.55

From: Jeff King <peff@peff.net>
Date: Feb 11, 2023, 03:00 UTC
Message-ID: <Y+cEwLjdgThXdkrw@coredump.intra.peff.net>
In-Reply-To: <xmqqilg9exq4.fsf@gitster.g>

On Fri, Feb 10, 2023 at 06:34:43PM -0800, Junio C Hamano wrote:

Show 11 quoted lines

> Jeff King <peff@peff.net> writes:
> 
> > I'm not sure this is really sufficient. As I was running t5559 over and
> > over to test various versions, I noticed a few other cases that seemed
> > to fail, some of them racily. So I'm a bit worried that the problem may
> > be more extensive, and it is only that this particular test happened to
> > trigger it reliably.
> 
> Yeah, I've seen not just t5559.30 (which is what the patch was
> about) reliably fail, but t5559.25 fail only from time to time.
> There may be others.

Ah, then if it's not just me, maybe it is time to take action. :)

I'm not sure if we have a complete list of good/bad versions, though. I know 2.0.10 was bad, and 2.0.9 seemed to be OK for me. 2.0.12 passes for me in regular use, but --stress seems to quickly barf. But it does for v2.0.9, too.

Show 5 quoted lines

> > In which case the right solution may be ditching t5559, or at the very
> > least adding a knob or version check to disable it.
> 
> Yup.  Do we have an example of checking versions of mod_frob being
> used from within our test suite?  I didn't locate one.

No, I don't think we do. The answers on stack overflow suggest using "strings" on the .so file. Not only is that gross, it doesn't seem to work in this case (I find "mod_http2 (v%s)", which is useless).

We could check the outer apache version with "apachectl -v", which is enough for Debian, but I think Todd had cases where the system was using a separate version of the mod, not the one that ships with apache itself.

I didn't see anything with IfVersion or IfModule that would let us depend on the module version. It does look like we can probably crank the LogLevel up and then grep for "mod_http2 (v[0-9.]+" in the log after starting the server. That would let us bail before actually running any tests.

We can be a bit cavalier here with technique, I think, because the worst case is that we say "this is not a known-good version" and just skip the http/2 tests.

Show 5 quoted lines

> > So I'm tempted to just punt for now and see how often this bites people
> > in the real world. And if it does become a problem, we'll have more data
> > on what exactly should be changed.
> 
> OK, works for me.

Well, I was all set to try a version check, but after seeing the race problems above, it doesn't seem like a good direction, at least not yet. So I'm content to sit on my hands for a while longer.

-Peff