Thread: 002_pg_upgrade is broken for custom install
Hi All, When oldinstall and olddump environment variables are set, 002_pg_upgrade.pl test does not finish and fails. # initializing database system by running initdb ok 1 - check locales in original cluster ok 2 - loaded old dump file # initializing database system by running initdb ok 3 # skip source node not using default install ok 4 - dump before running pg_upgrade # test failed stderr: # Tests were run but no plan was declared and done_testing() was not seen. # Looks like your test exited with 29 just after 4. Doesn't look like it's expected. Even if I rollback the history as far as f52345995d3691de7b0a888903b6cfed1497c672, the failure still appears. IIRC, the test was working fine till October 2024. But I might be wrong. Don't have energy to try git blame right now, but might do once I get some time. -- Best Wishes, Ashutosh Bapat
On 2025-Apr-04, Ashutosh Bapat wrote: > # test failed > stderr: > # Tests were run but no plan was declared and done_testing() was not seen. > # Looks like your test exited with 29 just after 4. > > Doesn't look like it's expected. Nope. Please have a look at the tmp_check/log/regress_002* log file -- that might contain a better indication of what the problem is. -- Álvaro Herrera 48°01'N 7°57'E — https://www.EnterpriseDB.com/
On Fri, Apr 4, 2025 at 5:02 PM Álvaro Herrera <alvherre@alvh.no-ip.org> wrote:
>
> On 2025-Apr-04, Ashutosh Bapat wrote:
>
> > # test failed
> > stderr:
> > # Tests were run but no plan was declared and done_testing() was not seen.
> > # Looks like your test exited with 29 just after 4.
> >
> > Doesn't look like it's expected.
>
> Nope. Please have a look at the tmp_check/log/regress_002* log file --
> that might contain a better indication of what the problem is.
Not much there
[17:08:28.162](0.939s) ok 3 # skip source node not using default install
# Running: pg_dumpall --no-sync --dbname port=13779
host=/tmp/LiPa_UJpSb dbname='postgres' --file
/home/ashutosh/work/units/pghead/build/dev/testrun/pg_upgrade/002_pg_upgrade/data/tmp_test_X1bh/dump1.sql
[17:08:30.129](1.967s) ok 4 - dump before running pg_upgrade
connection error: 'psql: error: connection to server on socket
"/tmp/LiPa_UJpSb/.s.PGSQL.13779" failed: FATAL: database "regression"
does not exist'
while running '/home/ashutosh/work/units/pghead/build/dev/bin/psql
--no-psqlrc --no-align --tuples-only --quiet --dbname port=13779
host=/tmp/LiPa_UJpSb dbname='regression' --file - --variable
ON_ERROR_STOP=1' at
/home/ashutosh/work/units/pghead/coderoot/pg/src/test/perl/PostgreSQL/Test/Cluster.pm
line 2256.
# Postmaster PID for node "old_node" is 779230
### Stopping node "old_node" using mode immediate
# Running: pg_ctl --pgdata
/home/ashutosh/work/units/pghead/build/dev/testrun/pg_upgrade/002_pg_upgrade/data/t_002_pg_upgrade_old_node_data/pgdata
--mode immediate stop
waiting for server to shut down.... done
server stopped
# No postmaster PID for node "old_node"
# No postmaster PID for node "new_node"
# No postmaster PID for node "dst_node"
[17:08:30.241](0.112s) # Tests were run but no plan was declared and
done_testing() was not seen.
[17:08:30.241](0.000s) # Looks like your test exited with 29 just after 4.
Something is happening at line 492
command_checks_all(
[
'pg_upgrade', '--no-sync',
'--old-datadir' => $oldnode->data_dir,
'--new-datadir' => $newnode->data_dir,
'--old-bindir' => $oldbindir . '/does/not/exist/',
'--new-bindir' => $newbindir,
'--socketdir' => $newnode->host,
'--old-port' => $oldnode->port,
'--new-port' => $newnode->port,
$mode, '--check',
],
1,
[qr{check for ".*?does/not/exist" failed}],
[],
'run of pg_upgrade --check for new instance with incorrect binary path');
because I don't see that test name or the next one being printed in
regress_*.log
--
Best Wishes,
Ashutosh Bapat
On 2025-Apr-04, Ashutosh Bapat wrote:
> connection error: 'psql: error: connection to server on socket
> "/tmp/LiPa_UJpSb/.s.PGSQL.13779" failed: FATAL: database "regression"
> does not exist'
> while running '/home/ashutosh/work/units/pghead/build/dev/bin/psql
> --no-psqlrc --no-align --tuples-only --quiet --dbname port=13779
> host=/tmp/LiPa_UJpSb dbname='regression' --file - --variable
> ON_ERROR_STOP=1' at
> /home/ashutosh/work/units/pghead/coderoot/pg/src/test/perl/PostgreSQL/Test/Cluster.pm
> line 2256.
> # Postmaster PID for node "old_node" is 779230
> ### Stopping node "old_node" using mode immediate
This is saying that Cluster->psql() (line 2256) tried to connect to
database regression and failed unexpectedly. Is your "olddump" file
created with pg_dumpall from a cluster that contains such a database?
If not, then the test isn't broken, you're just not operating it
correctly :-)
Maybe the failure could be clearer: rather than blindly trying to
connect to regression, first see if it exists, and die with a hard
failure ("your old dump must contain database regression") if not.
Alternatively, if the regression database doesn't exist, just skip that
part.
--
Álvaro Herrera Breisgau, Deutschland — https://www.EnterpriseDB.com/
On Fri, Apr 4, 2025 at 5:23 PM Álvaro Herrera <alvherre@alvh.no-ip.org> wrote:
>
> On 2025-Apr-04, Ashutosh Bapat wrote:
>
> > connection error: 'psql: error: connection to server on socket
> > "/tmp/LiPa_UJpSb/.s.PGSQL.13779" failed: FATAL: database "regression"
> > does not exist'
> > while running '/home/ashutosh/work/units/pghead/build/dev/bin/psql
> > --no-psqlrc --no-align --tuples-only --quiet --dbname port=13779
> > host=/tmp/LiPa_UJpSb dbname='regression' --file - --variable
> > ON_ERROR_STOP=1' at
> > /home/ashutosh/work/units/pghead/coderoot/pg/src/test/perl/PostgreSQL/Test/Cluster.pm
> > line 2256.
> > # Postmaster PID for node "old_node" is 779230
> > ### Stopping node "old_node" using mode immediate
>
> This is saying that Cluster->psql() (line 2256) tried to connect to
> database regression and failed unexpectedly. Is your "olddump" file
> created with pg_dumpall from a cluster that contains such a database?
> If not, then the test isn't broken, you're just not operating it
> correctly :-)
My bad, didn't pay attention to that error (even while pasting it) :(.
You are right. I used pg_dump (without --create) to create olddump.
The comment about olddump just mentions dump output. I think we should
mention pg_dumpall there.
>
> Maybe the failure could be clearer: rather than blindly trying to
> connect to regression, first see if it exists, and die with a hard
> failure ("your old dump must contain database regression") if not.
> Alternatively, if the regression database doesn't exist, just skip that
> part.
Right. Something like attached?
--
Best Wishes,
Ashutosh Bapat