Re: Effects of IDLE processes

From: Christopher Browne <cbbrowne(at)acm(dot)org>
To: pgsql-performance(at)postgresql(dot)org
Subject: Re: Effects of IDLE processes
Date: 2005-02-21 03:18:31
Message-ID: m3r7jaa9dk.fsf@knuth.knuth.cbbrowne.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

After a long battle with technology, Gaetano Mendola <mendola(at)bigfoot(dot)com>, an earthling, wrote:
> JM wrote:
>> Hi ALL,
>>
>> I was wondering if there is a DB performance reduction if
>> there are a lot of IDLE processes.
>>
>> 30786 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 32504 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 32596 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 1722 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 1724 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 3881 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 6332 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 6678 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 6700 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 6768 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 8544 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 8873 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 8986 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 9000 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 9010 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 9013 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 9016 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 9019 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>> 9020 ? S 0:00 postgres: user1 gmadb 10.10.10.1 idle
>>

> In my experience not at all, you have to wonder if some of that are
> "idle in transaction" that are really a pain in the @#$

I'd be concerned about "idle" processes insofar as they are holding on
to _some_ memory that isn't shared.

"idle in transaction" is quite another matter; long-running
transactions certainly do lead to evil. When running Slony-I, for
instance, "idle in transaction" means that pg_listener entries are
being held onto so they cannot be vacuumed out, and that's only one
example of a possible evil...
--
(reverse (concatenate 'string "moc.liamg" "@" "enworbbc"))
http://linuxdatabases.info/info/languages.html
You know how most packages say "Open here". What is the protocol if
the package says, "Open somewhere else"?

In response to

Responses

Browse pgsql-performance by date

  From Date Subject
Next Message Tom Lane 2005-02-21 03:48:06 Re: bad performances using hashjoin
Previous Message David Brown 2005-02-21 03:01:20 Re: bad performances using hashjoin