Horton

Re: reducing isolation tests runtime

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Alvaro Herrera <alvherre@2ndquadrant.com>, Thomas Munro <thomas.munro@enterprisedb.com>, pgsql-hackers@postgresql.org
Date: 2019-02-13T17:03:39Z
Lists: pgsql-hackers 
On 2019-02-13 10:58:50 -0500, Tom Lane wrote:
> Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> > On 2019-Feb-13, Tom Lane wrote:
> >> Some of the slower buildfarm critters use MAX_CONNECTIONS to limit
> >> the load on their hosts.  As long as the isolation tests honor that,
> >> I don't see a real need for a separate serial schedule.
> 
> > MAX_CONNECTIONS was the only reason I didn't push this through.  Do you
> > (Andres) have any solution to that?
> 
> Doesn't the common pg_regress.c infrastructure handle that?
> We might need to improve isolation_main.c and/or the isolation
> Makefile to make it accessible.

> I suppose that in what I'm thinking about, MAX_CONNECTIONS would be
> interpreted as "max number of concurrent isolation scripts", which
> is not exactly number of connections.  A quick and dirty answer
> would be to have isolation_main.c divide the limit by a factor of 4
> or so.

I guess that could work, although it's certainly not too pretty.
Alternatively we could pre-parse the spec files, but that's a bit
annoying given isolationtester.c is a separate c file...

Do you have an idea why we have both max_concurrent_tests *and*
max_connections in pg_regress? ISTM the former isn't really useful given
the latter?

Greetings,

Andres Freund

Commits

  1. Update obsolete sentence in README.parallel.

  2. Rewrite ConditionVariableBroadcast() to avoid live-lock.

  3. Tweak parallel hash join test case in hopes of improving stability.

  4. Rename pg_rewind's copy_file_range() to avoid conflict with new linux syscall.

  5. Fix some minor errors in new PHJ code.

  6. Fix EXPLAIN ANALYZE output for Parallel Hash.

  7. Fix rare assertion failure in parallel hash join.

  8. Cancel CV sleep during subtransaction abort.

  9. Add parallel-aware hash joins.

  10. Fix EXPLAIN ANALYZE of hash join when the leader doesn't participate.

  11. Add some regression tests that exercise hash join code.