- For all stemcells (Python source implementation)
- Automatically defines tags based on deployments, names and jobs
- Process, network, ntp and disk integrations by default
- Monit processes are added automatically to process integration
- You can define additional integrations
Upload the release to Bosh director
Create a runtime-config.yaml
file:
releases:
- name: datadog-agent
version: 1
addons:
- name: dd-agent
jobs:
- name: dd-agent
release: datadog-agent
properties:
dd:
use_dogstatsd: yes
api_key: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
tags: ["pe", "bosh", "bosh:bosh-exp"]
tags:
owner: pe
email: [email protected]
Upload runtime-config to Bosh Director: bosh update runtime-config runtime-config.yaml
Re-deploy the deployments to automatically add the agent.
You can also add the release to the manifest and deploy it
After cloning this repository, run:
git submodule init
git submodule update
to update the gohai submodule. Gohai is a binary (golang) program used by Datadog Agent to collect
basic information from the server. There is a script bosh_prepare
which also runs the git
submouldes commands and downloads all sources to the blobs
directory:
./bosh_prepare
It will get all sources specified in the packages/*/spec files (commented out) for local development, then you can create the release (final) with:
./bosh_final_release
which will upload the tarball to github and make the new blobs public available in the S3 bucket
bosh_prepare
ensures that gohai
git submodule is populated in src/gohai
. In order
to avoid including a git package -only needed because of go get
command to download
the dependencies- the script runs go get
locally to get all dependencies, so it is
assuming you have go
installed. After those dependencies were downloaded, the
binary package can be compiled, but a new folder src/gohai/src
appears, because of
that bosh will complain about untracked git changes, but there are two workarounds:
- use
bosh create release --force
- add to the
.gitmodules
file a config parameter for ignoring dirty state of the submodule:ignore = dirty
Here, the second option is used.
Springer Nature Platform Engineering, Jose Riguera Lopez ([email protected])
Copyright 2017 Springer Nature
Apache 2.0 License