Re: Addition of authenticated ID to pg_stat_activity

From: Michael Paquier <michael(at)paquier(dot)xyz>
To: Stephen Frost <sfrost(at)snowman(dot)net>
Cc: Andres Freund <andres(at)anarazel(dot)de>, Postgres hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Addition of authenticated ID to pg_stat_activity
Date: 2021-04-27 00:59:18
Message-ID: YIdh5j7yY0Z40mh3@paquier.xyz
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Apr 26, 2021 at 03:21:46PM -0400, Stephen Frost wrote:
> * Andres Freund (andres(at)anarazel(dot)de) wrote:
>> I'm getting a bit worried about the incremental increase in
>> pg_stat_activity width - it's probably by far the view that's most
>> viewed interactively. I think we need to be careful not to add too niche
>> things to it. This is especially true for columns that may be wider.
>>
>> It'd be bad for discoverability, but perhaps something like this, that's
>> not that likely to be used interactively, would be better done as a
>> separate function that would need to be used explicitly?
>
> I mean.. we already have separate functions and views for this, though
> they're auth-method-specific currently, but also provide more details,
> since it isn't actually a "one size fits all" kind of thing like this
> entire approach is imagining it to be.

Referring to pg_stat_ssl and pg_stat_gssapi here, right? Yes, that
would be very limited as this leads to no visibility for LDAP, all
password-based authentications and more.

I am wondering if we should take this as an occasion to move some data
out of pg_stat_activity into a separate biew, dedicated to the data
related to the connection that remains set to the same value for the
duration of a backend's life, as of the following set:
- the backend PID
- client_addr
- client_hostname
- client_port
- authenticated ID
- application_name? (well, this one could change on reload, so I am
lying).

It would be tempting to move the database name and the username but
these are popular fields with monitoring. Maybe we could name that
pg_stat_connection_status, pg_stat_auth_status or just
pg_stat_connection?
--
Michael

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Masahiro Ikeda 2021-04-27 01:03:20 Re: Transactions involving multiple postgres foreign servers, take 2
Previous Message Bharath Rupireddy 2021-04-27 00:53:56 Re: Enhanced error message to include hint messages for redundant options error