Docker Desktop Alternatives

Halil Kaskavalci — Sat, 13 Nov 2021 10:39:01 +0000

Docker recently updated its license. This made Docker for Desktop a paid product for enterprises. In this post, you will learn about your alternatives.

What does Docker Desktop do?

Docker is for Linux systems. In order to run containers, you need a Linux operating system. If you use MacOS or Windows, then you will need a Linux VM in order to run Docker commands. This is where Docker Desktop comes in.

Docker desktop is an end to end solution that includes a VM, Linux distribution and docker daemon that runs inside it. Your host machine communicates with the VM to run docker commands in it.

What are the Docker Desktop alternatives?

There are couple of alternatives if you want to get rid of Docker Desktop for any reason. Some are mature, some are still in early stages.

Podman

Podman is a daemonless container engine for developing, managing, and running OCI Containers on your Linux System. It is open source. It is a docker replacement. Podman is compatible with Docker API and CLI. You can switch as easy as alias docker=podman.

However, podman still lacks some features such as supporting Docker socket by default or compose features. Community is actively working to fill the gaps.

Installation guide (Mac only)

# Install podman
$ brew install podman
# Download VM
$ podman machine init
# Start the VM
$ podman machine start
# Start using podman. (save this to your shell profile)
$ alias docker=podman

Support for applications that need to use docker socket (DOCKER_HOST). See podman issue #11397.

# Create a SSH tunnel to the guest OS
$ ssh -nNT -L/tmp/podman.sock:/run/user/1000/podman/podman.sock -i ~/.ssh/podman-machine-default ssh://core@localhost:55591
# Use podman as DOCKER_HOST
$ export DOCKER_HOST='unix:///tmp/podman.sock'

Minikube

minikube is a managed VM to run Kubernetes on your local machine. It is possible to configure it to spit out environment configuration for Docker and Podman.

You will need minikube to be running in order to use Docker commands.

# Install minikube
$ brew install minikube
# Start local cluster
$ minikube start
# Consume docker env
$ eval $(minikube docker-env)

Note that you will run a complete kubernetes cluster to run Docker commands. This may or may not be an issue for your setup.

Rancher Desktop

Rancher Desktop is another alternative to Docker Desktop. It is maintained by Rancher. In my experience I wasn't able to fully integrate it. It is under active development. It is definitely an alternative.

Conclusion

I have been using podman actively for a week now. In my experience, podman is the slimmest and stable alternative to Docker Desktop.

Upgrade OpenTelemetry Go Instrumentation Libraries in Microservices

Halil Kaskavalci — Sat, 23 Oct 2021 07:29:00 +0000

OpenTelemetry is the work-in-progress merge of OpenCensus and OpenTracing. It is still in early stages as of 2021. We are using its instrumentation libraries opentelemetry-go and [opentelemetry-go-contrib]https://github.com/open-telemetry/opentelemetry-go-contrib/) for our Go services.

This post is about how to upgrade your instrumentation libraries in a micro service environment. You guessed it, it's not straightforward.

Upgrade is easy, just run go get

Yes and no. Opentelemtry-go is still actively being developed and is not General Release yet. So, expect breaking changes. Some packages may be renamed or disappear completely. When you upgrade, you need to be aware of breaking changes and stop using the deprecated packages or methods.

You can either review Changelog which offers a great history of features, bug fixes and breaking changes. PR numbers for every breaking change are given next to the line. For instance:

Metric SDK/API implementation type InstrumentKind moves into sdkapi sub-package. (#2091)

Just search for PR #2091 in the repository and find out what has changed. Within that PR maintainers should also fix all references for this breaking change in the codebase so you know what to change to adopt the breaking change.

Sometimes though, there are too many breaking change you can follow. Then, I recommend checking their examples. In example folder there are small examples how you can use certain functionalities. Such as, exporting to Prometheus. Verify if what you are doing in your code matches with the steps there.

Upgrade is done, what's with microservices?

If you have a single service that does not depend on others, you can skip that part. Imagine the following scenario.

In this dependency graph, Service A and Service B both use otel v0.20.0. Service B depends on Service A. Now, Service B wants to upgrade to otel v0.25.0. That may not work. Even though every dependency in Service B is now compatible with otel v0.25.0, since Service B depends on Service A, go mod will still download otel v0.20.0 artifacts. When there are breaking changes, you will see an error of ambiguous import like this:

$ go mod tidy
github.com/your-service imports
        go.opentelemetry.io/contrib/instrumentation/github.com/gin-gonic/gin/otelgin tested by
        go.opentelemetry.io/contrib/instrumentation/github.com/gin-gonic/gin/otelgin.test imports
        go.opentelemetry.io/contrib/propagators/b3: ambiguous import: found package go.opentelemetry.io/contrib/propagators/b3 in multiple modules:
        go.opentelemetry.io/contrib/propagators v0.20.0 (~/workspace/gow/pkg/mod/go.opentelemetry.io/contrib/propagators@v0.20.0/b3)
        go.opentelemetry.io/contrib/propagators/b3 v1.0.0 (~/workspace/gow/pkg/mod/go.opentelemetry.io/contrib/propagators/b3@v1.0.0)

There will be no references to v0.20.0 in your go.mod file since you are not using it. However, in go.sum you can find the hashes of the previous version.

Just tell me how to solve it already

Upgrade with respect to your dependency graph. In our scenario, you will need to upgrade otel in Service A first. Then, upgrade Service A dependency in Service B. That will likely give otel dependency errors to you. Upgrade otel dependencies and resolve breaking changes.

Congratulations, you have upgraded your instrumentation libraries.

DEV Community: Halil Kaskavalci