Frequently Asked Questions¶

Why does Percona Agent stop on its own?
How to install Percona Agent on Amazon RDS?
How to install Percona Agent on a server with old password authentication?
How to set up Percona Agent to use a proxy server?
Which IP addresses does Percona Agent use?
Does Percona Agent store data locally?
What do broken pipe log messages mean?

Why does Percona Agent stop on its own?¶

If you are not stopping the percona-agent process manually, it is likely that the API is sending a stop command. We regularly release new versions of Percona Agent with support for new features in PCT. Older versions may become deprecated due to lack of necessary functionality. In this case, we program the API to stop agents older than a specific version.

Make sure to regularly update as described in Updating Percona Agent.

The current minimum supported version is 1.0.11.

How to install Percona Agent on Amazon RDS?¶

This is currently not possible, because Percona Agent does not support monitoring multiple remote MySQL instances. It is a planned feature, which will be available soon.

How to install Percona Agent on a server with old password authentication?¶

If you need to install Percona Agent on a server with old password authentication, specify the -old-passwords=true option.

For MySQL 4.1 and later versions, you may have old passwords enabled explicitely. In this case, you can disable old passwords for one session while you install Percona Agent:

mysql> SET SESSION old_passwords=0;

Then create the percona-agent user and install Percona agent with the -create-mysql-user=false option.

How to set up Percona Agent to use a proxy server?¶

Percona Agent does not support proxies. It is a planned feature, which will be available soon for HTTPS proxies.

Which IP addresses does Percona Agent use?¶

Percona Agent communicates using the URI specified by the link entry in /usr/local/percona/percona-agent/config/agent.conf.

Does Percona Agent store data locally?¶

Percona Agent aggregates collected data in the /usr/local/percona/percona-agent/data folder. It usually reports collected data in regular intervals, but if there is a connection issue, this data is stored until the agent manages to send it.

As of Percona Agent 1.0.13, the following limits apply to local data:

1 hour
10 MiB
100 files

Previous versions did not limit the size of the data folder, which lead to situations when Percona Agent filled up disk space indefinitely.

What do `broken pipe` log messages mean?¶

Messages in percona-agent.log are not necessarily from Percona Agent. This file is used to capture many various messages.

You should refer to the agent log in the web UI for specific Percona Agent errors. To view the agent log, open Configure > Agent, and click info for the agent you want to troubleshoot. For more information, see Agent Configuration.

The write unix /var/lib/mysql/mysql.sock: broken pipe message is generated by the Go MySQL driver. It indicates a connection issue, and possibly a problem with credentials for the percona-agent user to access MySQL.

First of all, make sure that there is a percona-agent user set up for the MySQL instance. If it is, try to restart the agent by running the following command:

$ service percona-agent restart

or

$ /etc/init.d/percona-agent restart