datree VS reviewdog

Hi HN, I’m Shimon, the co-founder of Datree: A policy management solution for Kubernetes. We help DevOps engineers prevent misconfigurations in their Kubernetes by enforcing an organizational policy on their clusters. Engineers can define a custom policy or use one of Datree’s built-in policies, such as NIST/NSA Hardening Guide, EKS Security Best Practices, CIS Benchmark, and more.

Our website is at https://datree.io and our GitHub is here: https://github.com/datreeio/datree

This is not the first time I have shown Datree to the HN community: A little over a year ago, I posted here an earlier version of Datree (https://news.ycombinator.com/item?id=28918850). At that time, Datree consisted of a CLI tool to detect Kubernetes misconfigurations during the development process (locally or in the CI/CD), unlike the version I present today in which the enforcement happens in production.

We built the CLI tool because we detected a big problem among Kubernetes operators: Misconfigurations. Kubernetes is extremely complex and flexible, which makes it very easy to poorly configure it in ways that are not secure. And indeed, we talked to dozens of Kubernetes operators who suffered from various problems, starting with failed audits, all the way to downtime in production, all because of misconfigurations.

Our solution was simple: Give the developers the means to shift-left security testing during the development process with a CLI tool that can be integrated into the CI/CD. We thought this was the best way to approach the problem: It is easiest to fix misconfigurations in the development process before they are deployed to production, it prevents context-switching and relieves resources from the DevOps team.

While the CLI tool was very popular among the open-source community (it got over 6000 stars on GitHub), we soon realized that CI/CD enforcement is not enough. As we talked with Datree’s users, we realized we had made a fundamental mistake: We thought of misconfiguration prevention in technical terms rather than organizational terms.

Indeed, from a technical point of view, it makes sense to shift-left Kubernetes security. But when considering the organizational structure in which it takes place, it simply isn’t enough. DevOps engineers told us that they love the shift-left concept, but they simply cannot rely on the goodwill of the engineers to run a CLI tool locally or to monitor all the pipelines leading to production. They need governance, something to help them stay in control of the state of their clusters.

Moreover, we realized that many companies who use Kubernetes are heavily regulated, and cannot take any chances with their security. Sure, these companies want the engineers to fix misconfigurations during development, but they also want something to make sure that no matter what, their clusters remain misconfiguration-free.

Based on this understanding, we developed a new version of Datree that sits on the cluster itself (rather than in the CI/CD) and protects the production environment by blocking misconfigured resources with an admission webhook. It has a centralized policy management solution to enable governance, and native monitoring to get real-time insights into the state of your Kubernetes.

I look forward to hearing your feedback and answering any questions you may have.

Go is a statically typed, compiled programming language designed at Google, it is syntactically similar to C, but with memory safety, garbage collection, structural typing, and CSP-style concurrency. In my case, I needed to run Go for JSON schema validations, in other cases, you might want to perform a CPU-intensive task or use a CLI tool written in Go.

I worked with both of them and from my experience Gatekeeper is more solid and accountable, I even wrote an article about Gatekeeper. Both Gatekeeper and Kyverno require a lot of heavy lifting work. On the one hand, Gatekeeper will probably require more configuration work however the community and the tool itself are more stable than Kyverno. On the other hand, Kyverno policy-as-code capabilities are much easier to use/understand. This way or another, for me using Kyverno’s policy language or Rego for my policies, wasn’t such a pleasant experience. I personally believe in GitOps and shifting left so if you’re looking for tools I would highly recommend you to review Datree, which is an open-source CLI (Disclaimer: I’m one of the developers at Datree). Datree is a more centralized policy management solution rather than a policy engine. Unlike Kyverno/Gatekeeper Datree was built to help DevOps teams to shift left and practice GitOps by delegating more responsibilities to the developers more efficiently. In practice, Datree already comes with built-in rules and policies along with YAML and schema validation for K8s resources and CRDs such as Argo CRDs. Datree’s policies are written in JSONScheme which is a common solid policy language supported by the community for many years. Additionally, Datree’s CLI also comes with a dashboard app where you can monitor the policies in your organization. You can modify and update your policies, review which policies are being used in practice, and control who can create/delete/update your policies. The major difference is that at the moment, unlike Kyverno/Gatekeeper Datree doesn’t provide native policy enforcement in the Kubernetes cluster at the moment but we expect to release this support very soon. At the moment, we provide a way to scan the cluster using a kubectl plugin. Feel free to check it out :)

$ helm plugin install https://github.com/datreeio/helm-datree Installing helm-datree... https://github.com/datreeio/datree/releases/download/1.0.6/datree-cli_1.0.6_Darwin_x86_64.zip % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 673 100 673 0 0 1439 0 --:--:-- --:--:-- --:--:-- 1469 100 6901k 100 6901k 0 0 1852k 0 0:00:03 0:00:03 --:--:-- 2865k helm-datree is installed. See https://hub.datree.io for help getting started. Installed plugin: datree

GitHub

I build a general converter from SARIF to Reviewdog Diagnostic Format (RDFormat), then use Reviewdog to give suggested code changes as well as the context of the changes for PR reviewing.

I helped improve the Kubescape GitHub Actions fix suggestions code review process, where I created the workflow which works by collecting the SARIF (Static Analysis Results Interchange Format) file that kubescape generates. Then, with the help of HollowMan6/sarif4reviewdog, convert the SARIF file into RDFormat (Reviewdog Diagnostic Format) and generate reviews for code fix suggestions on GitHub Actions using Reviewdog. I also helped add the “fix" object support for the Kubescape-generated SARIF report.

Static analysis should be performed continuously by CI (Continuous Integration) to prevent unintentional inclusion of code. The JSON output schema is compatible with reviewdog and can be used as is in reviewdog.

Trigger reviewdog checks on a repository via GitLab webhook rather than CI job.

# parameters - op: add path: /spec/params/- value: name: report-file default: reportfile description: Report file with errors - op: add path: /spec/params/- value: name: format default: golint description: Format of error input from the task - op: add path: /spec/params/- value: name: reporter default: local description: Reporter type for reviewdog https://github.com/reviewdog/reviewdog#reporters - op: add path: /spec/params/- value: name: diff default: git diff FETCH_HEAD description: Diff command https://github.com/reviewdog/reviewdog#reporters # workspaces - op: add path: /spec/workspaces/- value: name: token description: | Workspace which contains a token file for Github Pull Request comments. Must have a token file with the Github API access token # steps - op: add path: /spec/steps/- value: name: reviewdog-report image: golangci/golangci-lint:v1.31-alpine # both have the same workspace name workingDir: $(workspaces.source.path) script: | #!/bin/sh set -ue wget -O - -q https://raw.githubusercontent.com/reviewdog/reviewdog/master/install.sh | sh -s -- -b $(go env GOPATH)/bin export REVIEWDOG_GITHUB_API_TOKEN=$(cat $(workspaces.token.path)/token) cat $(params.reportfile) | reviewdog -f=$(params.format) -diff="$(params.diff)"

In reference article, use prettier to format the markdown and the code snippets. I implement a text review using textlint and reviewdog in addition to that.

JetBrains Qodana is one option, but currently requires glue code to map the findings to MR comments. I'm using reviewdog for it but I'm hoping they'll eventually fix it to have native integration

I'm trying to make a GitHub action which automatically runs tsc to find TypeScript errors. Those errors should be annotated inline in the PR/Commits. I found reviewdog, which should work perfectly for this - but I played around with that for about 4 hours now and can't seem to get it to report errors successfully. tsc exits with code 2, but reviewdog still says that everything went fine. So I'm trying to find another solution for this, has anyone here done this before? For comparison, I managed to do the same thing with ESLint by adding a custom formatter to the eslint command (-f param),‍ which then gets automatically picked up by the GitHub action - but I can't find something similar for tsc..

I used ReviewDog to wire in Qodana results, so I hear you about wishing it was built in, but it is achievable: https://github.com/reviewdog/reviewdog#reporter-gitlab-merge...

Based on my contact with GitLab's built-in other scanning tools, I wouldn't trust their vuln management further than I could throw it, so you're likely not missing much on that front