Weaveworks shutdown: Impact on GitOps?

While our book is still in print, we have to digest the sad news that Weavworks shuts down commercial operations.

What does that mean for the contents of our book, and much more important, what does that mean for GitOps as such?

Future of Flux

The biggest impact surely is on the Flux project. First off, Flux is an open source project hosted by the CNCF. So the project itself remains as is. However, as stated in our book, at the time of writing more than two thirds of all contributions to the project originate from Weaveworks.

Fortunately, several former employees of Weaveworks (like Flux maintainer Stefan Prodan and Soule Ba) found a new home at ControlPlane. ControlPlane also now offer an “Enterprise Distribution for Flux CD”. In addition, there are other vendors that integrated Flux into their products like Azure, GitLab, AWS, GiantSwarm or Nutanix (which recently acquired D2iQ). These affirmed their support for Flux.

With all of this, maintenance of the core Flux project seems secured. Still, it remains to be seen if the number of full-time maintainers remain constant for the upstream Flux project.

Moreover, the future of the downstream Weaveworks products is uncertain. This includes primarily Weave GitOps (the open source project, including the UI and GitOps Run) and its enterprise version (formerly closed source, including GitOpsSet controller). The good news is that Weaveworks (as a kind of last act of their open source spirit) published the source code of Weave GitOps Enterprise. So, after a possible license change the then open source project could be maintained by others.

Before that happens, though, it’s unlikely to be of much use to the broader public as enterprises are unlikely to pick an unmaintained project that does not offer a binary distribution with unclear licensing for their production workloads.

Other projects that were backed by Weaveworks are Flamingo, or Flux Subsystem for Argo (FSA) as well as the terraform controller. While flamingo already had its own open source organization at GitHub, the terraform controller now moved from Weaveworks/tf-controller to flux-iac/tofu-controller. There, it is maintained by some of their former maintainers (like Chanwit Kaewkasi), now no longer employed by Weaveworks.

Future of GitOps

While the above provides updates to most of the references we had to Weaveworks in our book and prints a somewhat optimistic picture for the future of the Flux project, one question remains:

What does the unravelling of Weaveworks mean for GitOps as a concept?

We would like to put another question first: does the business side of one company developing a popular product reflect on the tool itself or even the idea it implements? There are other examples that proof otherwise, especially in open source like Docker, Sun Microsystems (Java, MySQL, etc.) or Hortonworks (Hadoop).

So we tend to think that the shutdown of Weavework’s commercial open core model is not related to Flux or the ideas behind GitOps.

Still, without Weaveworks the innovations created by the Flux project might be slowed down.

However, even if this should lead to a standstill of the project (which seems unlikely at the moment), other options remain. GitOps as a concept is implemented by other tools such as the CNCF projects Argo CD and pipecd, Rancher’s Fleet or JenkinsX.

One thing that will take time to get accustomed to is the fact that the Alexis Richardson’s famous blog post that coined the term GitOps is now only available via archive.org.

Conclusion

Summing up, we’re still convinced GitOps is here to stay. It might have lost the company that first came up with the idea, and maybe one of its most charismatic leaders. But the idea has spread far beyond it, is implemented by several tools and used in production widely. So the worst thing we expect is a temporary slowdown of innovations.

UPDATES:

• 2024-03-20: Add link to CNCF article, which allows for more optimism
• 2024-03-22: Add link to ControlPlane blog