pg_upgrade: Skip tests for live server checks on Windows
The tests added by 88aa6ca138a1 could fail on Windows, due to the fact
that pg_ctl is not able to compare the PID stored in postmaster.pid with
the postmaster that has been spawned. The postmaster may not be seen as
running, and could fail the live check.
Note that pg_ctl falls back on checking the start time recorded in
postmaster.pid, with a lag of 2 seconds. So, outside the possibility of
fixing pg_ctl so as the PID file checks become more reliable, an
alternative solution would be to use the same trick as the pg_ctl test
001_start_stop.pl, with a hardcoded sleep of *3* seconds, at least. I
cannot get excited about spreading this pattern, though, so skip it is
as we do not lose any coverage.
Per report from buildfarm members fairywren.
Discussion: https://postgr.es/m/apphYnTYS1WtLkKU@paquier.xyz
Branch
------
master
Details
-------
https://git.postgresql.org/pg/commitdiff/ac1ccbea98f166548890f14c9ca9d4ab055ca291
Modified Files
--------------
src/bin/pg_upgrade/t/002_pg_upgrade.pl | 73 +++++++++++++++++++---------------
1 file changed, 40 insertions(+), 33 deletions(-)
