Horton

Re: Avoid erroring out when unable to remove or parse logical rewrite files to save checkpoint work

Nathan Bossart <nathandbossart@gmail.com>

View thread View on postgr.es
From: Nathan Bossart <nathandbossart@gmail.com>
To: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>
Cc: Thomas Munro <thomas.munro@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Andres Freund <andres@anarazel.de>, "Bossart, Nathan" <bossartn@amazon.com>, Julien Rouhaud <rjuju123@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-08-23T18:07:44Z
Lists: pgsql-hackers 
On Wed, Aug 17, 2022 at 12:39:26PM +0530, Bharath Rupireddy wrote:
> +	/*
> +	 * We're only handling directories here, skip if it's not ours. Also, skip
> +	 * if the caller has already performed this check.
> +	 */
> +	if (!slot_dir_checked &&
> +		lstat(path, &statbuf) == 0 && !S_ISDIR(statbuf.st_mode))
>  		return;

There was previous discussion about removing this stanza from
ReorderBufferCleanupSeralizedTXNs() completely [0].  If that is still
possible, I think I would choose that over having callers specify whether
to do the directory check.

> +		/* we're only handling directories here, skip if it's not one */
> +		sprintf(path, "pg_replslot/%s", logical_de->d_name);
> +		if (get_dirent_type(path, logical_de, false, LOG) != PGFILETYPE_DIR)
> +			continue;

IIUC an error in get_dirent_type() could cause slots to be skipped here,
which is a behavior change.

[0] https://postgr.es/m/20220329224832.GA560657%40nathanxps13

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com

Commits

  1. Expand the use of get_dirent_type(), shaving a few calls to stat()/lstat()

  2. fsync pg_logical/mappings in CheckPointLogicalRewriteHeap().

  3. Introduce logical decoding.