Re: [PATCH 3/3] iotests: check: multiprocessing support

[Kevin Wolf]

Am 10.12.2021 um 15:46 hat Vladimir Sementsov-Ogievskiy geschrieben:
> 10.12.2021 17:36, Kevin Wolf wrote:
> > Am 03.12.2021 um 13:22 hat Vladimir Sementsov-Ogievskiy geschrieben:
> > > Add -j <JOBS> parameter, to run tests in several jobs simultaneously.
> > > For realization - simply utilize multiprocessing.Pool class.
> > > 
> > > Notes:
> > > 
> > > 1. Of course, tests can't run simultaneously in same TEST_DIR. So,
> > >     use subdirectories TEST_DIR/testname/ and SOCK_DIR/testname/
> > >     instead of simply TEST_DIR and SOCK_DIR
> > > 
> > > 2. multiprocessing.Pool.starmap function doesn't support passing
> > >     context managers, so we can't simply pass "self". Happily, we need
> > >     self only for read-only access, and it just works if it is defined
> > >     in global space. So, add a temporary link TestRunner.shared_self
> > >     during run_tests().
> > > 
> > > Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> > 
> > Just wondering, is it worth even supporting the mp=false case or can we
> > simplify the code a bit by always going through multiprocessing and
> > using nice directory names even if only one process is spawned?
> > 
> > Maybe John's observation that directory names get longer might be a
> > reason not to do that by default. Any other reasons you're aware of?
> 
> I just wanted to keep the behavior without a new option unchanged, to
> not deal with possible CI failures on "make check": who know what
> multiprocessing brings together with performance.

So basically just err on the side of caution. Makes sense.

Kevin