Analyze database connection usage
A “top”-like tool that monitors database connections across many processes and servers. The initial use case is that of tracking the large numbers of Python processes and database connections used by Openstack, however the system works with any SQLAlchemy application.
First, install connmon with pip.
Connmon has a config file which by default is in /etc/connmon.cfg. A simple file looks like:
# sample config, listen on 0.0.0.0:5800 and clients # will connect to 192.168.1.205:5800 [connmon_service_default] name: default nodes: node1 hostname=192.168.1.205:5800
This file is consulted by the connmon daemon as well as all client connections in order to establish in what location(s) the daemon is running.
If testing only on a single host, connmon can be used without a config file; if the file isn’t present, a default config that looks like the following is used:
# default config if no /etc/connmon.cfg and no # --config <file> option is passed [connmon_service_default] name: default nodes: node1 hostname=localhost:5800 bind=0.0.0.0
Then, the connmon daemon may be started:
This daemon listens for clients which will give it information about how many connections they’re using. It stores this state in memory and can then report on it. If the daemon is stopped or becomes unavailable to clients, each client keeps track of its state locally and will continue to try and reconnect to the daemon. When it does, it will bring the daemon up to date with its connection status.
To configure a stats client, connmon provides a plugin to the SQLAlchemy engine. To enable usage of the plugin in an application that talks to the database, use a URL like this:
The URL format above will get a little simpler in SQLAlchemy 1.1 where we’ll add some more portable ways to bundle “plugins” with a database URL.
Finally, we can view current connections using the console:
Then start up the applications that use the database. Everyone with the config will establish a TCP connection to the console at “192.168.1.205:5800”.
See the screenshot.png included.
Connecting without Config
The URL can also specify a specific host/port:
Configuring an HA Cluster
Multiple nodes can be configured to each run connmon listen such that they form a cluster; start up a server on each node with connmond --node <nodename>:
[connmon_service_default] name: default nodes: node1 hostname=192.168.1.205:5800 node2 hostname=192.168.1.206:5800 node3 hostname=192.168.1.207:5800
In the above model, all nodes contact each other and share all events. Connecting to the “default” servicename without a node name will cause the client to connect to a random node in the cloud. It will try each node until it finds one that connects.
Using With Devstack
Here are the magic incantations to add connmon to a devstack setup. First install connmon globally. Then in local.conf (note the double slash escapes):
[[post-config|$NOVA_CONF]] [database] connection = mysql+pymysql_connmon://root:firstname.lastname@example.org/nova?charset=utf8\\&connmon_service=default [api_database] connection = mysql+pymysql_connmon://root:email@example.com/nova_api?charset=utf8\\&connmon_service=default [[post-config|$NEUTRON_CONF]] [database] connection = mysql+pymysql_connmon://root:firstname.lastname@example.org/neutron?charset=utf8\\&connmon_service=default [[post-config|$KEYSTONE_CONF]] [database] connection = mysql+pymysql_connmon://root:email@example.com/keystone?charset=utf8\\&connmon_service=default [[post-config|$CINDER_CONF]] [database] connection = mysql+pymysql_connmon://root:firstname.lastname@example.org/cinder?charset=utf8\\&connmon_service=default [[post-config|$GLANCE_API_CONF]] [database] connection = mysql+pymysql_connmon://root:email@example.com/glance?charset=utf8\\&connmon_service=default
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size connmon-0.3.1.tar.gz (13.6 kB)||File type Source||Python version None||Upload date||Hashes View|