Обсуждение: TAP tests of pg_rewind not stopping servers used for the tests

Hi all,

In the refactoring of pg_rewind tests committed as 53ba107, it happens
that some of the servers used for the tests are not stopped at the end
of the test. The issue is that RewindTest.pm uses END to stop the
servers, but now that the local and remote tests are grouped half of
the servers are not stopped.

The reason why I did not notice that before is because as pg_rewind
uses local Unix socket to work on the nodes, the tests are not
failing.
Sorry about that, that's my fault. And attached is a patch fixing the
issue: it replaces END by a cleanup function called at the end of each
local/remote test to be sure that the servers are shut down.
Regards,
--
Michael

On 04/15/2015 03:07 PM, Michael Paquier wrote:
> Hi all,
>
> In the refactoring of pg_rewind tests committed as 53ba107, it happens
> that some of the servers used for the tests are not stopped at the end
> of the test. The issue is that RewindTest.pm uses END to stop the
> servers, but now that the local and remote tests are grouped half of
> the servers are not stopped.
>
> The reason why I did not notice that before is because as pg_rewind
> uses local Unix socket to work on the nodes, the tests are not
> failing.
> Sorry about that, that's my fault. And attached is a patch fixing the
> issue: it replaces END by a cleanup function called at the end of each
> local/remote test to be sure that the servers are shut down.

Thanks, committed. I kept the END block, though, so that we still clean 
up if the test dies with an exception.

- Heikki

On Thu, Apr 16, 2015 at 1:57 AM, Heikki Linnakangas wrote:
> Thanks, committed. I kept the END block, though, so that we still clean up
> if the test dies with an exception.

That makes sense. Thanks.
-- 
Michael