Re: [HACKERS] Incomplete startup packet errors

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Andrew Dunstan <andrew(dot)dunstan(at)2ndquadrant(dot)com>
Cc: Christoph Berg <myon(at)debian(dot)org>, pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: Re: [HACKERS] Incomplete startup packet errors
Date: 2019-03-03 20:52:31
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-hackers

I wrote:
> Andrew Dunstan <andrew(dot)dunstan(at)2ndquadrant(dot)com> writes:
>> Patch proposed by Christoph Berg is here:

> Meh. That doesn't silence only the zero-bytes case, and I'm also
> rather afraid of the fact that it's changing COMMERROR to something
> else. I wonder whether (if client_min_messages <= DEBUG1) it could
> result in trying to send the error message to the already-lost
> connection. It might be that that can't happen, but I think a fair
> amount of rather subtle (and breakable) analysis may be needed.

Concretely, what about doing the following instead? This doesn't provide
any mechanism for the DBA to adjust the logging behavior; but reducing
log_min_messages to DEBUG1 would not be a very pleasant way to monitor for
zero-data connections either, so I'm not that fussed about just dropping
the message period for that case. I kind of like that we no longer need
the weird special case for SSLdone.

regards, tom lane

Attachment Content-Type Size
alternate-fix-for-incomplete-packet-logging.patch text/x-diff 1.7 KB

In response to


Browse pgsql-hackers by date

  From Date Subject
Next Message Ramanarayana 2019-03-03 20:55:35 Re: psql show URL with help
Previous Message Pavel Stehule 2019-03-03 20:27:13 Re: [HACKERS] proposal: schema variables