Horton

Re: [HACKERS] Incomplete startup packet errors

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Jobin Augustine <jobinau@gmail.com>
Cc: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>, Robert Haas <robertmhaas@gmail.com>, Christoph Berg <myon@debian.org>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2019-11-25T15:02:42Z
Lists: pgsql-hackers 
Jobin Augustine <jobinau@gmail.com> writes:
> However, Checking whether the port is open is resulting in error log like:
> 2019-11-25 14:03:44.414 IST [14475] LOG:  invalid length of startup packet
> Yes, This is different from "Incomplete startup packet" discussed here.

> Steps to reproduce:
> $ telnet localhost 5432
>> 
>> 

Well, the agreed-to behavior change was to not log anything if the
connection is closed without any data having been sent.  If the
client *does* send something, and it doesn't look like a valid
connection request, I think we absolutely should log that.

			regards, tom lane

Commits

  1. Don't log incomplete startup packet if it's empty