The LINBIT® team has been busy for more than a year on development work that could put DRBD® 9 in the mainline Linux kernel, possibly as soon as the Linux 7.2 release sometime in September or October 2026. The current version of the DRBD kernel module that is in the mainline Linux kernel is 8.4.11.
While the work to put DRBD 9 into the upstream Linux kernel is not top secret, admittedly the LINBIT team did not promote its efforts widely in the early or even mid-stages. Now that a Phoronix article has scooped LINBIT on the story, it seems necessary to talk more about this development work and what it means to DRBD and Linux users.
To summarize the significant scope of work involved in replacing the current DRBD 8.4.11 version in the Linux kernel with a 9.3 version, LINBIT CEO and DRBD creator, Philipp Reisner said in an Insights from Phil newsletter:
The most recent release was 8.4.11, eight years ago. However, the development of [DRBD 9] started in October 2011. It was like a rewrite, in place. Over the course of the journey, we had to touch on every line of code to introduce the concept that a cluster can have multiple peers.
Differences between DRBD 8 and DRBD 9
Indeed, a major difference between the two DRBD versions is that the older 8.4 version was intended for 2-node failover clusters. DRBD 9 supports failover clusters up to 31 peers for each DRBD-replicated volume. While this makes possible some extreme high availability (HA), most HA clusters will use three to five nodes.
DRBD 9 also includes the concept of quorum that works in clusters of three or more nodes. This DRBD quorum feature helps prevent data divergence, so-called split-brain scenarios, when nodes unexpectedly leave the cluster, for example, due to a node-level failure. Other nodes that have maintained quorum will essentially mark the failed node as unable to promote to a primary role when it returns to the cluster, until after it has synchronized with a known good data set from peers holding quorum.
Without DRBD quorum, 2-node DRBD cluster administrators needed to resort to using fencing or STONITH solutions to avoid data divergence, and manual intervention was needed upon a failed node returning to the cluster.
You can read more about this and other DRBD 9 features in the DRBD 9 User Guide.
Preparing to upgrade from DRBD 8 to DRBD 9
If you are currently administering an HA cluster running DRBD 8.4 from the mainline Linux kernel, you can learn more about the process of upgrading to DRBD 9 by reading Upgrading DRBD from 8.4 to 9.x in the DRBD 9 User Guide. While the process has many steps, it is straightforward and many systems administrators have done it already. If you think you might need help or have questions about the process, specific to your own production or enterprise environment, you can contact the LINBIT team to learn about support options.
Conclusion
You can expect to read and hear more from the LINBIT team about its progress to put DRBD 9 into the upstream Linux kernel in the coming months. This will include updates from LINBIT developers themselves. When the merge review dust settles and the hoped-for acceptance of DRBD 9 into the mainline Linux kernel happens, it will be a positive event for existing DRBD 8.4 and DRBD 9 users. Through the merge review process, LINBIT developers, including Christoph Böhmwalder, the LINBIT developer who leads the efforts, are already hardening the DRBD 9 code as it gets put through independent analysis and testing.
