Jia Tan owned the previous newsletter with their work on the XZ/ssh backdoor. I wanted to write about HashiCorp and Redis, and a common theme between the two companies of changing their licensing of important software projects from open-source licenses to proprietary licenses. Several news sites have featured articles related to the topic. Here is one opinion piece by Steven Vaughan-Nichols for Computerworld.
I understand his critique of venture capitalism. In these cases, startups opt for rapid growth through venture capital. They leverage open source to gain attention and followers, only to shift gears when investors prioritize profitability over idealism. It is an abuse of open-source, and our community’s defense is forking.
OpenTofu, Terraform’s fork, was quickly moved under the hood of the Linux Foundation, which was also directly involved in creating the Valkey fork of Redis.
Building open-source projects connected to a commercial entity that pays the developers’ salaries and is a reliable support provider for customers requires careful right-sizing of the entity. For me, organic growth works best for that.
At LINBIT, we are continuously improving our build automation. That includes fetching fresh kernel updates from our target Linux distributions, building and testing existing DRBD packages with those, and building fresh DRBD packets if necessary. Since kernels from a minor release are ABI compatible on the RHEL platform, we only have a single DRBD build for multiple kernels. kernels.drbd.io is a new live display of our automated testing results with distribution kernels. I hope that clarifies which DRBD builds fit with which red hat kernel. We will add more distributions over time.
The team has been productive on the blog recently. First, in a recent blog post titled, Disaster Recovery Is Not a Technology, It’s a Plan, LINBIT Solutions Architect Yusuf Yıldız shares his belief that aggressive marketing by software vendors has mistakenly led to the perception that DR is purely a technological solution implemented within storage or virtualization environments.
However, DR involves more than software alone – it is a comprehensive strategy that integrates diverse technologies, requiring businesses to recognize this broader perspective to mitigate potential risks and ensure effective recovery.
As explained in the blog post, Using the DRBD Quorum Feature as an Alternative to Fencing in a Pacemaker Cluster, “Implementing fencing is a way to ensure the consistency of your replicated data by avoiding ‘split-brain’ scenarios. When communication between cluster nodes breaks, fencing prevents the data from diverging among your data replicas.”
Using Fencing in Pacemaker Clusters on VirtualBox Hypervisors describes how you can configure the fence_vbox fence agent in high-availability (HA) Pacemaker development clusters running on VirtualBox for Linux. Read the post to learn how to use fencing in your development clusters and get some ideas on how you can try breaking your development clusters to test the fence agents of your cluster.
There have been several software releases since I last wrote, too. WinDRBD 1.1.12 is mainly for users that use fencing instead of quorum to avoid split brain. Three patches to DRBD and several small fixes are included in the WinDRBD code.
linstor-proxmox v8.0.0 contains two new features, which you can read about in the hyperlink. You need to know that this version requires LINSTOR 1.27.0 or later. Following this release, linstor-proxmox v8.0.1 is a small bugfix release that contains two changes related to cloud-init images.
linstor-gateway v1.5.0 was released with no further changes since the rc announcement the previous week. For the full changelog, you can see the release notes on GitHub.
Last but not least, I am aligning the DRBD releases 9.1.20 and 9.2.9 with the Ubuntu 24.04 (“noble”) release. The rc.1 is already out.
