Re: What is considered a "slow processes"

Lists: pgadmin-support
From: "Benjamin Krajmalnik" <kraj(at)servoyant(dot)com>
To: <pgadmin-support(at)postgresql(dot)org>
Subject: What is considered a "slow processes"
Date: 2011-01-06 16:53:18
Message-ID: F4E6A2751A2823418A21D4A160B689887B0C9D@fletch.stackdump.local
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgadmin-support

While playing around with pgadmin to see the state of running queries, I
noticed that all of the queries are falling under the category of "slow
processes" or "idle process". Some of the process whose query is
showing as idle are tagged as idle processes, while some are tagged as
slow processes.

Could this be an "artifact" of using a connection pooler, since the
client start times are sometimes a few minutes old? What is the
criteria for determining if a process is "slow"?

Running pg 8.4 on FreeBSD 7.2 amd64.


From: Guillaume Lelarge <guillaume(at)lelarge(dot)info>
To: Benjamin Krajmalnik <kraj(at)servoyant(dot)com>
Cc: pgadmin-support(at)postgresql(dot)org
Subject: Re: What is considered a "slow processes"
Date: 2011-01-09 22:52:53
Message-ID: 4D2A3C45.90102@lelarge.info
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgadmin-support

Le 06/01/2011 17:53, Benjamin Krajmalnik a écrit :
> While playing around with pgadmin to see the state of running queries, I
> noticed that all of the queries are falling under the category of "slow
> processes" or "idle process". Some of the process whose query is
> showing as idle are tagged as idle processes, while some are tagged as
> slow processes.
>

There is actually more possibilities:

* idle process: for each process that don't currently run a query and
aren't in a transaction block
* blocked process: for each process waiting on a lock
* active process: for each process executing a query since less than
10 seconds
* slow process: for each process executing a query since at least
10 seconds

> Could this be an "artifact" of using a connection pooler, since the
> client start times are sometimes a few minutes old? What is the
> criteria for determining if a process is "slow"?
>

We use query_start to check if a query is slow, so I don't think it has
anything to do with a connection pooler.

> Running pg 8.4 on FreeBSD 7.2 amd64.
>

--
Guillaume
http://www.postgresql.fr
http://dalibo.com