Message-ID: <013104103e016e84d59c1e2accb0b1b1f99d9870.camel@cybertec.at>
Subject: Re: Replication Testing- How to introduce a Lag
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: "Subramanian,Ramachandran" <ramachandran.subramanian@alte-leipziger.de>,
  "pgsql-novice@lists.postgresql.org"	 <pgsql-novice@lists.postgresql.org>
Date: Tue, 24 Mar 2026 08:07:00 +0100
In-Reply-To: <aa307627991c41f280bb6da27ebc9b1d@alte-leipziger.de>
References: <aadb9058660d4e10b5cee44b82458802@alte-leipziger.de>
					 <f3927bc3a9baf4400325b91652869907f0581c97.camel@cybertec.at>
					 <e1e41761323040228570cf6a7e02518e@alte-leipziger.de>
				 <8b38fa5141905f9ad6983b8d5ae944e812e3d28a.camel@cybertec.at>
				 <fa47fe18078043adbdfe007b7270ffd0@alte-leipziger.de>
			 <fc42f77c80fdb3c0267148b807edfacc49b0b0ca.camel@cybertec.at>
			 <fc3be1f56ded49dc9be519f92d8b9ab4@alte-leipziger.de>
		 <0e00fcc0062b59b88f98dba4900ce49d5a9a3fea.camel@cybertec.at>
		 <477369bf897b44feb2d2368f6fca93fc@alte-leipziger.de>
	 <89d63402747edb3bf80ff668fdd63cc77334f446.camel@cybertec.at>
	 <aa307627991c41f280bb6da27ebc9b1d@alte-leipziger.de>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.58.3 (3.58.3-1.fc43) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/013104103e016e84d59c1e2accb0b1b1f99d9870.camel%40cybertec.at>
Precedence: bulk

On Tue, 2026-03-24 at 06:14 +0000, Subramanian,Ramachandran wrote:
> I noticed that if I insert one row in a table at the source, the differen=
ce in LSNs
> is not 1 .=C2=A0 ( with a delibrately introduced delay on the apply side =
),=20
>=20
> It is sometimes 96, sometimes 296 ( for the same table two inserts=C2=A0 =
) .

Right, because the LSN is not a counter that increases with each new WAL
record.  It is a position in the WAL stream.  The difference between the
LSNs of two adjacent WAL records is not 1, but the byte count of the first
WAL record.

For example: if you insert a larger row, the LSN will advance more.
Note also that not all inserts will produce the same kind of WAL:
one insert might write a full page image to the WAL, while the next
a normal insert record.

> Is there a method to calculate the APPROXIMATE amount of data in ( Bytes =
)
> that are yet to be transfered from Source to Standby ?

That's exactly what pg_wal_lsn_diff() does.

What is your worry?  What is your ultimate goal?

Yours,
Laurenz Albe