Horton

Re: Decoding speculative insert with toast leaks memory

Dilip Kumar <dilipbalaut@gmail.com>

View thread View on postgr.es
From: Dilip Kumar <dilipbalaut@gmail.com>
To: Amit Kapila <amit.kapila16@gmail.com>
Cc: Andres Freund <andres@anarazel.de>, Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>, Peter Eisentraut <peter.eisentraut@2ndquadrant.com>, Petr Jelinek <petr.jelinek@enterprisedb.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Tomas Vondra <tomas.vondra@enterprisedb.com>
Date: 2021-06-07T13:15:04Z
Lists: pgsql-hackers 
On Mon, Jun 7, 2021 at 6:34 PM Amit Kapila <amit.kapila16@gmail.com> wrote:

> On Mon, Jun 7, 2021 at 6:04 PM Dilip Kumar <dilipbalaut@gmail.com> wrote:
> >
> > I have fixed all pending review comments and also added a test case
> which is working fine.
> >
>
> Few observations and questions on testcase:
> 1.
> +step "s1_lock_s2" { SELECT pg_advisory_lock(2); }
> +step "s1_lock_s3" { SELECT pg_advisory_lock(2); }
> +step "s1_session" { SET spec.session = 1; }
> +step "s1_begin" { BEGIN; }
> +step "s1_insert_tbl1" { INSERT INTO tbl1 VALUES(1, repeat('a', 4000))
> ON CONFLICT DO NOTHING; }
> +step "s1_abort" { ROLLBACK; }
> +step "s1_unlock_s2" { SELECT pg_advisory_unlock(2); }
> +step "s1_unlock_s3" { SELECT pg_advisory_unlock(2); }
>
> Here, s1_lock_s3 and s1_unlock_s3 uses 2 as identifier. Don't you need
> to use 3 in that part of the test?
>

Yeah this should be 3.


>
> 2. In the test, there seems to be an assumption that we can unlock s2
> and s3 one after another, and then both will start waiting on s-1 but
> isn't it possible that before s2 start waiting on s1, s3 completes its
> insertion and then s2 will never proceed for speculative insertion?
>

I agree, such race conditions are possible.  Currently, I am not able to
think what we can do here, but I will think more on this.

-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. Fix ABI break introduced by commit 4daa140a2f.

  2. Fix valgrind issue in pgoutput.c.

  3. Fix decoding of speculative aborts.

  4. pgoutput: Fix memory leak due to RelationSyncEntry.map.

  5. Fix initialization of RelationSyncEntry for streaming transactions.

  6. Implement streaming mode in ReorderBuffer.