RE: Verify or dump pgdata files

Hi Muhammad,

Thank you for responding to my query.

I don’t think checksum method will work in my situation because the source file would have changed by the time I run checksum. So, the checksum (or diff) will be different.
I am doing rsync of active cluster.

What I am looking for is physical (structural) verification of target file such as beginning block and end block, beginning of file and end of file etc. All I want to know is “completeness” and “readability” of the file by postgres.

Thanks.

From: Muhammad Ikram <me(dot)mikram(at)gmail(dot)com>
Sent: Sunday, October 20, 2024 10:34 PM
To: Murthy Nunna <mnunna(at)fnal(dot)gov>
Cc: pgsql-admin(at)postgresql(dot)org
Subject: Re: Verify or dump pgdata files

[EXTERNAL] – This message is from an external sender
Hi,

Could you use diff ? e.g.

diff -qr /var/lib/pgsql/15/data /tmp/data

Or

pg_checksums --verify --data-dir=/pgdata (for checksum enabled databases)

or do checksum on source and target and take diff between two checksum generated files e.g.

cd /pgdata
find . -type f -exec sha256sum {} \; > /tmp/source_checksums.txt

Regards,
Ikram

On Sun, Oct 20, 2024 at 4:45 AM Murthy Nunna <mnunna(at)fnal(dot)gov<mailto:mnunna(at)fnal(dot)gov>> wrote:
Hi,

We are currently using pg_waldump to verify integrity of WAL files after archiving them in external storage. If the WAL is incomplete pg_waldump returns error, which is great.

Similarly, I am wondering if there is any utility that can be used against data files in /pgdata. Just to verify after “rsync” of /pgdata.

Thanks in advance for your response.