From: Heikki Linnakangas <hlinnaka@iki.fi>
To: pgsql-committers@lists.postgresql.org
Subject: Re: pgsql: Replace polyphase merge algorithm with a simple balanced k-way m
Date: Mon, 18 Oct 2021 16:46:35 +0300
Message-ID: <45e2ee8f-205e-6150-938e-3eecd0ed2fef@iki.fi> (raw)
In-Reply-To: <E1mcRXp-0000e9-0Y@gemulon.postgresql.org>
References: <E1mcRXp-0000e9-0Y@gemulon.postgresql.org>
On 18/10/2021 15:15, Heikki Linnakangas wrote:
> Replace polyphase merge algorithm with a simple balanced k-way merge.
This patch made some buildfarm animals fail in the isolation tests.
'eelpout' seems to have ran out of disk space:
> +setup failed: ERROR: could not write to file "base/pgsql_tmp/pgsql_tmp915834.0.fileset/0.1": No space left on device
Maybe I was unlucky. However, 'francolin', 'desmoxytes', 'dragonet' and
'idiacanthus' also failed, in the same test:
> ==~_~===-=-===~_~== pgsql.build/src/test/isolation/output_iso/regression.diffs ==~_~===-=-===~_~==
> diff -U3 /mnt/resource/bf/build/francolin/HEAD/pgsql.build/../pgsql/src/test/isolation/expected/multiple-row-versions.out /mnt/resource/bf/build/francolin/HEAD/pgsql.build/src/test/isolation/output_iso/results/multiple-row-versions.out
> --- /mnt/resource/bf/build/francolin/HEAD/pgsql.build/../pgsql/src/test/isolation/expected/multiple-row-versions.out 2021-10-05 04:13:04.811944633 +0000
> +++ /mnt/resource/bf/build/francolin/HEAD/pgsql.build/src/test/isolation/output_iso/results/multiple-row-versions.out 2021-10-18 12:34:35.036779419 +0000
> @@ -2,10 +2,9 @@
>
> starting permutation: rx1 wx2 c2 wx3 ry3 wy4 rz4 c4 c3 wz1 c1
> step rx1: SELECT * FROM t WHERE id = 1000000;
> - id|txt
> --------+---
> -1000000|
> -(1 row)
> +id|txt
> +--+---
> +(0 rows)
>
> step wx2: UPDATE t SET txt = 'b' WHERE id = 1000000;
> step c2: COMMIT;
> @@ -26,5 +25,4 @@
> step c4: COMMIT;
> step c3: COMMIT;
> step wz1: UPDATE t SET txt = 'a' WHERE id = 1;
> -ERROR: could not serialize access due to read/write dependencies among transactions
> step c1: COMMIT;
> ==~_~===-=-===~_~== inst/logfile ==~_~===-=-===~_~==
The symptom is different, there is no "out of disk space" error, but it
kind of looks like an index was built incorrectly, because the simple
SELECT doesn't return the row that it should.
That sure smells like a bug in the new sorting code. I haven't been able
to reproduce that yet, but I'll keep trying. It's strange that this
doesn't happen in any other test, that is a very ordinary table with a
very ordinary index.
- Heikki
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients from your mail client; the link fills in
the headers below: reply via email
Some clients, webmail especially, drop In-Reply-To and References from
a link, and the reply then starts a new thread: use the mbox below.
To: pgsql-committers@postgresql.org
Cc: hlinnaka@iki.fi
Subject: Re: pgsql: Replace polyphase merge algorithm with a simple balanced k-way m
In-Reply-To: <45e2ee8f-205e-6150-938e-3eecd0ed2fef@iki.fi>
References: <E1mcRXp-0000e9-0Y@gemulon.postgresql.org> <45e2ee8f-205e-6150-938e-3eecd0ed2fef@iki.fi>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox