Horton

Re: Add LSN along with offset to error messages reported for WAL file read/write/validate header failures

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>
Cc: Kyotaro Horiguchi <horikyota.ntt@gmail.com>, orlovmg@gmail.com, alvherre@alvh.no-ip.org, nathandbossart@gmail.com, pgsql-hackers@lists.postgresql.org
Date: 2022-12-19T08:07:35Z
Lists: pgsql-hackers

Attachments

On Tue, Dec 13, 2022 at 09:32:19PM +0530, Bharath Rupireddy wrote:
> Okay, here's the v5 patch that I could come up with. It basically adds
> functions for dissecting WAL file names and computing offset from lsn.
> Thoughts?

I had a second look at that, and I still have mixed feelings about the
addition of the SQL function, no real objection about
pg_dissect_walfile_name().

I don't really think that we need a specific handling with a new
macro from xlog_internal.h that does its own parsing of the segment
number while XLogFromFileName() can do that based on the user input,
so I have simplified that.

A second thing is the TLI that had better be returned as int8 and not
int4 so as we don't have a negative number for a TLI higher than 2B in
a WAL segment name.
--
Michael

Commits

  1. Rename pg_dissect_walfile_name() to pg_split_walfile_name()

  2. Add pg_dissect_walfile_name()

  3. Add LSN location in some error messages related to WAL pages