From: sergey-krasnogorov (@sergey-krasnogorov) <noreply+sergey-krasnogorov@github.com>
To: postgresql-interfaces/psqlodbc <noreply+postgresql-interfaces-psqlodbc@github.com>
Subject: Re: [postgresql-interfaces/psqlodbc] issue #224: psqlODBC driver or MS System.Data.Odbc do not report connection lost
Date: Mon, 16 Nov 2026 07:25:56 +0000
Message-ID: <gh-postgresql-interfaces-psqlodbc-224-c6910184904@github.com> (raw)
In-Reply-To: <gh-postgresql-interfaces-psqlodbc-224@github.com>
References: <gh-postgresql-interfaces-psqlodbc-224@github.com>
Hi Dave,
Thanks for your comments and suggestions.
Please see attached logs. They cover three separate different SW runs/scenarios:
1. **error**
Scenario: start db service, check `pg_isready`, in a few seconds start `SELECT`ing.
Error (connection is down) caught at third attempt (third db service start).
Unfortunately, based on our app internal log, it is random: first `SELECT` might be Ok, even second Ok, then it lost connection.
Please note: `psqlodbc_EchoLoader_<pid>.log` even not written. I never saw it created in such a scenario.
2. **no-pooling-no-error**
Scenario: exactly same as **1. error** but with driver connection pooling is off.
Please note: `psqlodbc_EchoLoader_<pid>.log` is there regardless of how "fast" I started `SELECT`ing.
3. **no-rush-no-error**
Scenario: start db service, check `pg_isready`, **wait** for min 30s, then start `SELECT`ing.
Please note: `psqlodbc_EchoLoader_<pid>.log` of course is there.
Looking forward to your opinion on what it might be.
Thanks,
Sergey
[#224_logs_02.zip](https://github.com/user-attachments/files/34902200/224_logs_02.zip)