Manual Entries of commit_delay and commit_siblings are not complete. - Mailing list pgsql-docs

From Tianyin Xu
Subject Manual Entries of commit_delay and commit_siblings are not complete.
Date
Msg-id CABBDWweXsOJiMOBTsJ3K4Cs0E8J8YKQS3hWk6_3GR2UFOykPGA@mail.gmail.com
Whole thread Raw
Responses Manual Entries of commit_delay and commit_siblings are not complete.  (Tianyin Xu <tixu@cs.ucsd.edu>)
Re: Manual Entries of commit_delay and commit_siblings are not complete.  (Bruce Momjian <bruce@momjian.us>)
List pgsql-docs
Hi, pgsql-docs,

I was confused by the system behavior of PG's write ahead log, specifically the relation between commit_delaycommit_siblings, and fsync.

Basically according to the manual (http://www.postgresql.org/docs/9.1/static/runtime-config-wal.html), I know:

"commit_delay" specifies the delay before a transaction attempts to flush the WAL buffer out to disk.
"commit_siblings" specifies minimum number of concurrent open transactions to require before performing the commit_delay delay.

However, after looking at the code, I figured out these configuration parameters are also controlled by fsync.

The only usage of the two parameter I can find is:

/* src/backend/access/transam/xact.c */

static TransactionId
RecordTransactionCommit(void)
     ...
     if (CommitDelay > 0 && 
                  enableFsync &&
                  MinimumActiveBackends(CommitSiblings))
            pg_usleep(CommitDelay);

    ...
}

where, CommitDelay stores the parameter of "commit_delay", CommitSiblings stores the parameter of "commit_siblings", and enableFsync stores the parameter of "fsync".

It seems the two parameters (commit_delay and commit_siblings) only have effect when fsync is set to be *on*.

But, unfortunately, such behavior is not covered in the manual. The mistake I made is only set the two values but disabling fsync so the expected PG does not behave as I expected. I think it's definitely helpful to specify the relations between these configuration parameters either in the manual, or in the source code by giving warnings logs.

Thanks a lot!
Tianyin

pgsql-docs by date:

Previous
From: Tom Lane
Date:
Subject: Re: [BUGS] BUG #7543: Invalid table alias: DELETE FROM table *
Next
From: Tom Lane
Date:
Subject: Docs issue for CREATE SCHEMA ... AUTHORIZATION permissions