From: Rural Hunter <ruralhunter@gmail.com>
Cc: pgsql-performance@postgresql.org <pgsql-performance@postgresql.org>
Subject: Re: Very slow planning performance on partition table
Date: Wed, 30 Jul 2014 12:05:25 +0800
Message-ID: <53D86F05.5060007@gmail.com> (raw)
In-Reply-To: <CAMkU=1wHC_w6D2kc1V6y1PDkRxDV3B+1fJBk3Z2L=SprGw7omw@mail.gmail.com>
References: <53CFB6BC.4050606@gmail.com>
<25010.1406253203@sss.pgh.pa.us>
<53D26875.2080604@gmail.com>
<53D51E49.102@gmail.com>
<24564.1406478487@sss.pgh.pa.us>
<CAMkU=1ydmpt9XRMxt0sPNnQsXEoF_c7bgp2kHxtDbPNGg5Vj5w@mail.gmail.com>
<53D7597E.2080209@gmail.com>
<CAMkU=1wHC_w6D2kc1V6y1PDkRxDV3B+1fJBk3Z2L=SprGw7omw@mail.gmail.com>
List-Unsubscribe: <mailto:majordomo@postgresql.org?body=unsub%20pgsql-performance>
This happened again. This time I got the connection status(between
pgbouncer host to pgsql host) at postgresql side. When the problem
happens, the connection status is this:
ESTABLISHED: 188
CLOSE_WAIT: 116
The count of connections in CLOSE_WAIT is abnormal. Comparing with
normal situation, there is usually no close_wait connection. The
connection status sample is like this:
ESTABLISHED: 117
CLOSE_WAIT: 0
I have 4 users configured in pgbouncer and the pool_size is 50. So the
max number of connections from pgbouncer should be less than 200.
The connection spike happens very quickly. I created a script to check
the connections from pgbouncer. The script checks the connections from
pgbouncer every 5 mins. This is the log:
10:55:01 CST pgbouncer is healthy. connection count: 73
11:00:02 CST pgbouncer is healthy. connection count: 77
11:05:01 CST pgbouncer is healthy. connection count: 118
11:10:01 CST pgbouncer is healthy. connection count: 115
11:15:01 CST pgbouncer is healthy. connection count: 75
11:20:01 CST pgbouncer is healthy. connection count: 73
11:25:02 CST pgbouncer is healthy. connection count: 75
11:30:01 CST pgbouncer is healthy. connection count: 77
11:35:01 CST pgbouncer is healthy. connection count: 84
11:40:10 CST Problematic connection count: 292, will restart pgbouncer...
Now I suspect there is some network problem between the hosts of
pgbouncer and pgsql. Will check more.
--
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: ruralhunter@gmail.com
Subject: Re: Very slow planning performance on partition table
In-Reply-To: <53D86F05.5060007@gmail.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox