DataDog/datadog-unix-agent

Name: datadog-unix-agent

Owner: Datadog, Inc.

Description: Datadog Unix Agent

Created: 2018-01-16 18:38:33.0

Updated: 2018-05-14 14:05:16.0

Pushed: 2018-05-24 10:54:42.0

Homepage: null

Size: 176

Language: Python

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

Datadog Unix Agent

A full fledged Agent specifically designed for Unix-based systems.

Note: This agent is currently in beta for AIX. It has not been tested on other systems.

This agent targets miscellaneous Unix operating systems not supported by our currently available agents. To do so, it has been stripped of unnecessary bloat and cut around irrelevant use-cases given the target platforms, sometimes at the expense of certain features. To maximize portability the agent will attempt to reduce the number of non pure-python dependencies to a minimum, and rely on packages with native support for the targeted OSes (AIX currently).

AIX

Development Requirements

• yum (use script here: https://ftp.software.ibm.com/aix/freeSoftware/aixtoolbox/ezinstall/ppc/yum.sh)
• python2.7+ (should be installed with the yum script)
• python-tools (should be installed with the yum script)
• gcc (install with yum)
• virtualenv (recommended)

Running the agent

To keep everything nice and clean we will stick to a virtual environment.

install virtualenv

Now let's create a virtual environment (assuming you're in the repo root and running bash):

r venv
venv/bin/activate

Runtime requirements

Most requirements can be transparently installed using the bundled requirements file.

install -r requirements.txt

If you're on Linux, or FreeBSD (and many others), you're probably good to go. However on AIX you would need the XL compiler available on the machine to be able to build. XL is not free and so we will not assume it will be available on your LPAR. Fortunately there are ways around this.

Currently the following steps should be executed before attempting to install via the requirements file provided in the repository.

We are currently working on merging some changes upstream into psutil to avoid having to apply any patches manually, but until then please use the psutil.patch file provided in the patch/ subdir on the vanilla psutil:

Clone the psutil repository somewhere in your filesystem, keep the repositories separate, so clone this outside of datadog-unix-agent.

clone git@github.com:giampaolo/psutil.git
h -p0 < psutil.patch

Then build the python package:

gcc -lgcc"  LDSHARED="/opt/freeware/lib/python2.7/config/ld_so_aix gcc -bI:/opt/freeware/lib/python2.7/config/python.exp" CFLAGS="-fno-strict-aliasing -Wall  -Wstrict-prototypes -fPIC -O2" python setup.py bdist_wheel 
install dist/psutil-5.4.5-cp27-cp27-aix_6_1.whl  # or similar 

Now head back to the datadog-unix-agent repo and install the remaining dependencies:

install -r requirements.txt

At this point you should be ready to go!

Running the agent

To run the agent you may simply execute:

OG_LEVEL=debug DD_API_KEY=<api_key_here> ./agent.py start

You may also drop a configuration file datadog.yaml in the repo root (you may use the example file provided in the repo). Other more UNIX appropriate locations are supported (namely /etc/datadog-agent but packaging isn't currently available). When using a config file instead of environment variables to specify configuration params, launch like this:

ent.py start

Config file will be searched in this order:

• ./datadog.yaml
• /etc/datadog-agent/datadog.yaml

By default, the agent will run in the foreground.

There are also facilities to run the agent both daemonized and via a supervisor, this would probably be the preferred way to manage your daemons (agent and dogstatsd).

Happy DataDoggin'!