Re: Rewriting the test of pg_upgrade as a TAP test - take three - remastered set

Michael Paquier <michael@paquier.xyz>

From: Michael Paquier <michael@paquier.xyz>
To: Andres Freund <andres@anarazel.de>
Cc: Noah Misch <noah@leadboat.com>, Tom Lane <tgl@sss.pgh.pa.us>, Thomas Munro <thomas.munro@gmail.com>, Julien Rouhaud <rjuju123@gmail.com>, Andrew Dunstan <andrew@dunslane.net>, Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-06-04T03:35:45Z
Lists: pgsql-hackers
On Fri, Jun 03, 2022 at 12:53:18PM -0700, Andres Freund wrote:
> [...]

TRAP: FailedAssertion("AmIoWorkerProcess()", File: "xlog.c", Line:
4860, PID: 35325)

> regress_log_002_pg_upgrade.log includes all of 002_pg_upgrade_old_node.log and
> 002_pg_upgrade_new_node.log. The old node's log includes all pg_dump queries.

log_statement = all is the part biting here.  It does not seem like
we'd lose a lot of context even if this is made less verbose.

> Followed by many MB of diff due to
> 
> === diff of /Users/admin/pgsql/src/bin/pg_upgrade/tmp_check/tmp_test_Q7GQ/dump1.sql and /Users/admin/pgsql/src/bin/pg_upgrade/tmp_check/tmp_test_Q7GQ/dump2.sql
> === stdout ===

Something like 80~85% of the bloat comes from the diffs in your case.
Well, it is always possible to limit that to an arbitrary amount of
characters (say 50k~100k?) to still give some context, and dump the
whole in a different file outside the log/ path (aka tmp_check/), so
that the buildfarm would show a minimum amount of information, while
local failures would still have an access to everything.

Do you have any preferences?
--
Michael

Commits

  1. Improve and fix some issues in the TAP tests of pg_upgrade

  2. Fix several issues with the TAP tests of pg_upgrade

  3. Make upgradecheck a no-op in MSVC's vcregress.pl

  4. Switch the regression tests of pg_upgrade to use TAP tests

  5. Fix check for PGHOST[ADDR] in pg_upgrade with Windows and temporary paths

  6. Remove REGRESS_OUTPUTDIR environment variable.

  7. Add PostgreSQL::Test::Cluster::config_data()