Message-ID: <gh-pgjdbc-pgjdbc-3511-c2648160654@github.com>
From: "rdhzl (@rdhzl)" <noreply+rdhzl@github.com>
To: "pgjdbc/pgjdbc" <noreply+pgjdbc-pgjdbc@github.com>
Date: Mon, 10 Feb 2025 14:26:59 +0000
Subject: Re: [pgjdbc/pgjdbc] issue #3511: Performance Regression in JDBC Driver 42.7.5 - getCrossReference
In-Reply-To: <gh-pgjdbc-pgjdbc-3511@github.com>
References: <gh-pgjdbc-pgjdbc-3511@github.com>
List-Id: <gh-pgjdbc-pgjdbc.github.com>
X-GitHub-Author-Login: rdhzl
X-GitHub-Comment-Id: 2648160654
X-GitHub-Comment-Type: issue_comment
X-GitHub-Issue: 3511
X-GitHub-Repo: pgjdbc/pgjdbc
X-GitHub-Type: comment
X-GitHub-Url: https://github.com/pgjdbc/pgjdbc/issues/3511#issuecomment-2648160654
Content-Type: text/plain; charset=utf-8

Just to follow up on the above, commenting out the Liquibase preconditions where we check for index existence in the schema does seem to mitigate the performance impacts. Liquibase is deciding what mechanism to do those checks under the hood. In general those checks seem to be "expensive" relative to other DDL, but they became substantially moreso with the 42.7.5.