After speaking with Sean Murphy and Daniel Messer it was decided that this statement is inaccurate and affecting customer adoption of the product. Please remove it in current documentation and future documentation. 1.1.1. Before you begin Expanding the existing volumes to span across more nodes is not supported.
The following are 2 statements in the RHHI docs which seemingly negate the possibility of expanding the initial volume deployment. Red Hat Hyperconverged Infrastructure 1.1 - Maintaining Red Hat Hyperconverged Infrastructure *************** **************** CHAPTER 1. ADD COMPUTE AND STORAGE RESOURCES Red Hat Hyperconverged Infrastructure (RHHI) can be scaled in multiples of three nodes to a maximum of nine nodes. 1.1. SCALING RHHI DEPLOYMENTS 1.1.1. Before you begin Be aware that the only supported method of scaling Red Hat Hyperconverged Infrastructure (RHHI) is to create additional volumes that span the new nodes. Expanding the existing volumes to span across more nodes is not supported. Red Hat Hyperconverged Infrastructure - 1.1 - Deploying Red Hat Hyperconverged Infrastructure *************** **************** 3.6. SUPPORT LIMITATIONS One arbitrated replicated volume is supported as part of the initial deployment of Red Hat Hyperconverged Infrastructure (RHHI). Expanding this arbitrated replicated volume is not supported. Adding additional arbitrated replicated volumes is not supported. Arbitrated replicated volumes are currently supported only in a replica 3 arbiter 1 configuration.
I tested volume expansion via the RHVM GUI in my lab today and it worked as expected. I expanded a volume used for running VMs from 3 bricks to 6 bricks. The GUI presented a warning message about potential degraded performance, see below. I wonder if that is related to rebalance operations (which have to be triggered manually) I used version 4.2 of RHV and 3.3u1 of RHGS ==GUI popup warning== Add Bricks warning: Expanding gluster volume in a hyper-converged setup is not recommended as it could lead to degraded performance. To expand storage for cluster, it is advised to add additional gluster volumes. Are you sure you want to continue? ==END==
Sas, Sahina, I'm deferring to you on this. My understanding has always been that the initially created volumes cannot be expanded across new nodes, and that new volumes need to be created instead. Keep in mind that we document what QE tests, verifies, and recommends rather than what is technically possible, or we risk leading customers into bad configurations.
While volume expansion is possible - there were some corner cases where moving from replicate to distribute-replicate fails, and we do not support this mode of scaling. QE has not tested this. Adding needinfo on Krutika to link to the known bugs around this
https://bugzilla.redhat.com/show_bug.cgi?id=1463907 https://bugzilla.redhat.com/show_bug.cgi?id=1434653 https://bugzilla.redhat.com/show_bug.cgi?id=1530146 are the ones I am aware of. -Krutika
Bug 1632144 logged to track this feature from engineering.
Thanks Sahina. Closing NOTABUG for now, and will add this back to the docs queue when the engineering feature is ready.