kubeapply
![Go Report Card](https://goreportcard.com/badge/github.com/henvic/kubeapply)
kubeapply is a microservice for running kubectl apply
through a web API.
kubeapply makes it easier to use Kubernetes in a declarative manner while Kubernetes API still doesn't have an endpoint similar to kubectl apply
.
Server-side Apply enhancement workaround
kubectl apply
is a core part of the Kubernetes config workflow. However, its implementation is in the client-side of a CLI tool. As of February 2019, there is work in progress to migrate the functionality to the server-side.
This middleware is a workaround useful for using Kubernetes kubectl apply
over an HTTP connection while work on this integration is still in progress.
Dependencies
- Go ≥ 1.11 to generate the server binary.
- Kubernetes 1.10 or greater.
Commands
You might want to run cmd/server --help
to list the available options.
The environment variable DEBUG
sets the logging to debug mode.
kubectl
must be available on the machine.
A Docker image is publicly available as wedeploy/kubeapply.
Kubernetes cluster configurations are stored in the /configurations directory.
Security
It is unsafe to run this software unless you protect this service appropriately.
You must run it on an isolated machine with limited network connectivity.
Port 9000 (API) is only accessible from localhost, except in the Docker image.
Port 8081 (debugging tool) is enabled by default only for localhost.
For your safety, you must assume that anyone who can reach this middleware endpoints has total control over the machine it is running on. Reasons: cluster options, file-system access, etc.
To communicate with other machines outside of a trusted network use a secure layer and proper client and server authentication protocols.
Endpoints
/version
curl http://localhost:9000/version -v
returns the local kubectl
version.
/apply
You can use all flags available on kubectl apply
(including global ones).
{
"flags": {
"dry-run": true,
"output": "wide"
},
"files": {
"relative/path/cert.yaml": "apiVersion: ...",
"server.json": {"apiVersion": "..."}
}
}
You can use command
attribute to call another kubectl command.
Example: "command": "create"
calls kubectl create
.
A JSON object is returned containing the explanation of the executed command and its result.
Some details:
cmd_line
is the corresponding command you can copy and paste on a shell to execute the command yourself.
exit_code
is the process exit code.
dir
is the relative path to the stored configuration and logs.
stderr
is always a string.
stdout
is JSON body by default. For other output formats, it is returned as a string value.
Recordings and logs
Configurations requested are recorded on a directory inside configurations
named by the id of the request and organized by date. No rotation policy is in place.
You don't need to pass the --filename
flag as if no file is found on your YAML, --filename=./
and --recursive
are automatically set.
Run example with --dry-run:
curl -d @example.json -v -XPUT http://localhost:9000/apply -H "Content-Type: application/json" | jq
Flags
You don't need to prefix flags or shortcuts with --
or -
. You also can use numbers or booleans directly.
Example:
{
"flags": {
"--dry-run": true,
"timeout": "1m",
"R": true,
"f": "service.yaml"
}
}
This configuration is similar to kubectl apply --dry-run=true --timeout=1m -R -f=service.yaml
.
Contributing
You can get the latest source code with go get -u github.com/henvic/kubeapply
The following commands are available and require no arguments:
In lieu of a formal style guide, take care to maintain the existing coding style. Add unit tests for any new or changed functionality. Integration tests should be written as well.
Committing and pushing changes
The master branch of this repository on GitHub is protected:
- force-push is disabled
- tests MUST pass on Travis before merging changes to master
- branches MUST be up to date with master before merging
Keep your commits neat and well documented. Try to always rebase your changes before publishing them.
Maintaining code quality
goreportcard can be used online or locally to detect defects and static analysis results from tools with a great overview.
Using go test and go cover are essential to make sure your code is covered with unit tests.
Always run make test
before submitting changes.