NVMe queue count selection not possible with Trident
Applies to
- NetApp AFF-A90 and AFF-C80 nodes
- NetApp Data ONTAP 9.X
- NetApp Trident 25.10.0 and 26.06.0
- OpenShift 4.19.32
- NVMe/TCP protocol
Issue
- After adding new ontap nodes AFF-A90 to an exiting ontap cluster, pods fail to mount NVMe/TCP volumes on the new nodes, using Trident
- Pod events show errors:
MountVolume.MountDevice failed for volume "pvc-yyyyzzzz-xxxx-yyyy-zzzz-xyxyxyxyxy": rpc error: code=Internal desc=failed to stage volume: failed to connect subsystem nqn.1992-08.com.netapp:sn.zzzzzzzz-yyyy-xxxx-yxyx-xyxyxyxyxyxy:subsystem.hostname-xxxxxxx-yyyy to XX.YY.ZZ.YY; exit status 1; failed to connect subsystem nqn.1992-08.com.netapp:sn.zzzzzzzz-yyyy-xxxx-yxyx-xyxyxyxyxyxy:subsystem.hostname-xxxxxxx-yyyy - Trident logs show:
Failed to get subsystem: no subsystem paths found; returning minimal subsystem Error="no subsystem paths found"could not add new controller: failed to write to nvme-fabrics device
- Customer observed that NVMe connections succeed on AFF-C80 (2 queues), but fail on AFF-A90 (4 queues)
- NVMe/TCP mount failures occur with default queue count of 4 on AFF-A90
