Message-ID: <gh-pgjdbc-pgjdbc-2918-c3254148468@github.com>
From: "davecramer (@davecramer)" <noreply+davecramer@github.com>
To: "pgjdbc/pgjdbc" <noreply+pgjdbc-pgjdbc@github.com>
Date: Thu, 04 Sep 2025 15:03:26 +0000
Subject: Re: [pgjdbc/pgjdbc] issue #2918: Set defaultRowFetchSize=10000 by default to improve app stability
In-Reply-To: <gh-pgjdbc-pgjdbc-2918@github.com>
References: <gh-pgjdbc-pgjdbc-2918@github.com>
List-Id: <gh-pgjdbc-pgjdbc.github.com>
X-GitHub-Author-Login: davecramer
X-GitHub-Comment-Id: 3254148468
X-GitHub-Comment-Type: issue_comment
X-GitHub-Issue: 2918
X-GitHub-Repo: pgjdbc/pgjdbc
X-GitHub-Type: comment
X-GitHub-Url: https://github.com/pgjdbc/pgjdbc/issues/2918#issuecomment-3254148468
Content-Type: text/plain; charset=utf-8

> I'd rather have a "safer" value in the ballpark of 128, and if that's too low for you, 1000 max, but nowhere above that. Currently pgJDBC eats RAM like it's unlimited. I would like that just as much as you do, but it isn't ;)
> 
> Also, why not honor the fetchSize on every SELECT statement regardless of auto-commit ?
> 
> Auto-commit should only affect writing statements (INSERT/UPDATE/DELETE), not read (SELECT) statements, right ? Why is auto-commit even relevant to honoring the fetchSize in the first place ? I'm confused.

This is an artifact of MVCC. You need to be in a transaction to create a snapshot to make sure that the data is consistent between fetches