From: | Andres Freund <andres(at)anarazel(dot)de> |
---|---|
To: | Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> |
Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, Stephen Frost <sfrost(at)snowman(dot)net>, Don Seiler <don(at)seiler(dot)us>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
Subject: | Re: [PATCH] Include application_name in "connection authorized" log message |
Date: | 2018-06-22 19:38:07 |
Message-ID: | 20180622193807.6jcwp5mno2ujxzc4@alap3.anarazel.de |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 2018-06-22 15:26:06 -0400, Alvaro Herrera wrote:
> On 2018-Jun-22, Andres Freund wrote:
>
> > On 2018-06-22 12:16:18 -0400, Robert Haas wrote:
>
> > > OK, that makes more sense, but I'm still skeptical of adding a special
> > > case particularly for application_name.
> >
> > I think a fair argument could be made that you'd want to have
> > application_name logged exactly once, not in every line. Just to cope
> > with log volume. With decent log analysis tools once is enough.
>
> Seems harder than it sounds ... because if the user turns off
> log_connections then it's not longer in the log.
That's superuser only, so I really don't quite buy that argument.
> And what about the application changing it after the fact?
I don't think that matters in a lot of scenarios. Poolers are the big
exception to that, obviously.
> One idea would be to have a log line designed specifically to be
> printed once at connection start (if not log_connections) and then
> once immediately after it changes. Am I the only one for whom this
> sounds like overengineering?
Yea. I think on balance, I don't buy that it's worth the cost. But I
don't think it's a clear cut "you don't need this".
Greetings,
Andres Freund
From | Date | Subject | |
---|---|---|---|
Next Message | Peter Geoghegan | 2018-06-22 19:43:10 | Re: [WIP] [B-Tree] Retail IndexTuple deletion |
Previous Message | Alvaro Herrera | 2018-06-22 19:26:06 | Re: [PATCH] Include application_name in "connection authorized" log message |