pg_waldump test failure - Mailing list pgsql-hackers

From Andres Freund
Subject pg_waldump test failure
Date
Msg-id ufmxlls5wa35cbvmxzxmfsh5iphrjhobr7naoxth7la323c4tx@e4gglpoja5ay
Whole thread Raw
List pgsql-hackers
Hi,

I was looking at CI on postgres's github mirror and noticed this failure:
https://cirrus-ci.com/task/6719398406258688

https://api.cirrus-ci.com/v1/artifact/task/6719398406258688/testrun/build/testrun/pg_waldump/001_basic/log/regress_log_001_basic

https://api.cirrus-ci.com/v1/artifact/task/6719398406258688/testrun/build/testrun/pg_waldump/001_basic/log/001_basic_main.log

The specific failing test only has the following output:

[17:59:47.615](0.000s) not ok 66 - all output lines are FPW
[17:59:47.615](0.000s) #   Failed test 'all output lines are FPW'
#   at /Users/admin/pgsql/src/bin/pg_waldump/t/001_basic.pl line 294.
[17:59:47.615](0.000s) #          got: '1'
#     expected: '0'

Unfortunately that's not really enough information to debug why the test
failed, and since it doesn't seem to be a regularly occurring issue I can't
just add one-off debugging instrumentation.


It's possible that this is caused by an IPC::Run issue that was fixed by Noah,
but that hasn't yet been released and thus not been picked up by CI:
https://www.postgresql.org/message-id/20250216235843.7c.nmisch%40google.com


Even if it's that issue, I think this failure shows the test ought to be more
debuggable than it is.  I guess the most basic thing would be to log
pg_waldump as a "note"?


Greetings,

Andres Freund



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Get rid of WALBufMappingLock
Next
From: Tom Lane
Date:
Subject: Re: New buildfarm animals with FIPS mode enabled