This project provides Sensu Go Assets containing portable Python runtimes (for various platforms), based on the excellent pyenv project. In practice, this Python runtime asset should allow Python-based scripts (e.g. Sensu Community plugins) to be packaged as separate assets containing Python scripts and any corresponding Python module dependencies. In this way, a single shared Python runtime may be delivered to systems running the new Sensu Go Agent via the new Sensu's new Asset framework (i.e. avoiding solutions that would require a Python runtime to be redundantly packaged with every python-based plugin).
Currently this repository only supports a subset of Linux distribution by making use of Docker containers to build and test. If you would like extend the coverage, please take a look at the travisCI integration and test build scripts. We're happy to take pull requests that extending the platform coverage. Here's the current platform matrix that we are testing for as of the 0.1 release:
Asset Platform | Tested Operating Systems Docker Images |
---|---|
alpine (based on alpine:3.8) | Alpine(3, 3.8, latest) |
centos6 (based on centos:6) | Centos(6, 7), Debian(8, 9, 10), Ubuntu(20.04, 16.04, 18.04 |
centos7 (based on centos:7) | Centos(7), Debian(8, 9, 10), Ubuntu(20.04, 16.04, 18.04 |
debian8 (based on debian:8) | Debian(8, 9, 10), Ubuntu(14.04, 16.04, 18.04), Centos(7,8) |
Please note that when using the Python runtime asset built on a target OS that is different from the build platform, you may need to explicitly set the SSL_CERT_DIR environment variable to match the target OS filesystem. Example: CentOS configures it libssl libraries to look for certs by default in /etc/pki/tls/certs
and Debian/Ubuntu use /usr/lib/ssl/certs
. The CentOS runtime asset when used on a Debian system would require the use of SSL_CERT_DIR override in the check command to correctly set the cert path to /usr/lib/ssl/certs
Please note the following instructions:
-
Use a Docker container to install
pyenv
, build a Python, and generate a local_build Sensu Go Asset.$ docker build --build-arg "PYTHON_VERSION=3.6.11" -t sensu-python-runtime:3.6.11-alpine -f Dockerfile.alpine . $ docker build --build-arg "PYTHON_VERSION=3.6.11" -t sensu-python-runtime:3.6.11-debian8 -f Dockerfile.debian8 .
-
Extract your new sensu-python asset, and get the SHA-512 hash for your Sensu asset!
$ mkdir assets $ docker run -v "$PWD/assets:/assets" sensu-python-runtime:3.6.11-debian8 cp /assets/sensu-python-runtime_3.6.11_debian8_linux_amd64.tar.gz /assets/ $ shasum -a 512 assets/sensu-python-runtime_3.6.11_debian8_linux_amd64.tar.gz
-
Put that asset somewhere that your Sensu agent can fetch it. Perhaps add it to the Bonsai asset index!
-
Create an asset resource in Sensu Go.
First, create a configuration file called
sensu-python-runtime-3.6.11-debian.json
with the following contents:{ "type": "Asset", "api_version": "core/v2", "metadata": { "name": "sensu-python-runtime-3.6.11-debian", "namespace": "default", "labels": {}, "annotations": {} }, "spec": { "url": "http://your-asset-server-here/assets/sensu-python-runtime-3.6.11-debian8.tar.gz", "sha512": "XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX", "filters": [ "entity.system.os == 'linux'", "entity.system.arch == 'amd64'", "entity.system.platform_family == 'debian'" ] } }
Then create the asset via:
$ sensuctl create -f sensu-python-runtime-3.6.11-debian.json
-
Create a second asset containing a Python script.
To run a simple test using the Python runtime asset, create another asset called
helloworld-v0.1.tar.gz
with a simple Python script atbin/helloworld.py
; e.g.:#!/usr/bin/env python from datetime import datetime now = datetime.now() current_time = now.strftime("%H:%M:%S") print("Hellow world! The current Time is ", current_time)
NOTE: this is a simple "hello world" example, but it shows that we have support for basic modules!
Compress this file into a g-zipped tarball and register this asset with Sensu, and then you're all ready to run some tests!
-
Create a check resource in Sensu Go.
First, create a configuration file called
helloworld.json
with the following contents:{ "type": "CheckConfig", "api_version": "core/v2", "metadata": { "name": "helloworld", "namespace": "default", "labels": {}, "annotations": {} }, "spec": { "command": "helloworld.py", "runtime_assets": ["sensu-python-runtime-3.6.11-debian", "helloworld-v0.1"], "publish": true, "interval": 10, "subscriptions": ["docker"] } }
Then create the asset via:
$ sensuctl create -f helloworld.json
At this point, the
sensu-backend
should begin publishing your check request. Anysensu-agent
member of the "docker" subscription should receive the request, fetch the Python runtime and helloworld assets, unpack them, and successfully execute thehelloworld.py
command by resolving the Python shebang (#!/usr/bin/env python
) to the Python runtime on the Sensu agent$PATH
.:wq
The Python runtime includes a basic set of standard python modules. If you want to use a python script that requires additional modules, you can package those additional modules with your script in an asset. However you will need to use a wrapper script that set the python module search path correctly. Please take a look at packaging python modules for detailed instructions on steps to take.