Having defined two PowerStore arrays in the secret with different blockProtocol, the wrong protocol is selected during PVC creation.
When the first array is set to NVMeTCP and the second to ISCSI the pvc, even when the storage class explicitly refers to the first one, attempt to connect to the volume via iSCSI which is not working since only NVMe host definitions have been created on the array.
None of the appliances is set as "default".
Logs
{"level":"info","msg":"/csi.v1.Node/NodeStageVolume: REP 0009: rpc error: code = Internal desc = unable to find device after multiple discovery attempts: discovery complete but devices not found","time":"2024-09-25T11:54:28.889578221Z"}
Screenshots
No response
Additional Environment Information
No response
Steps to Reproduce
Environment preparation
2 PowerStore arrays in secret. First one with NVMeTCP and second one with iSCSI
No running pods, pvc, pv
Remove any existing host definitions for worker nodes on both
Verify host definitions have been created with specified protocol on both the arrays.
Create workload which uses NVMeTCP storage class
oc create -f workload_nvmetcp.yaml
Verify pods, pvc and pv are created
oc get pod,pvc,pv -n
Result: pods are not being created as the CSI driver attempts to use iSCSI instead of NVMeTCP
Expected Behavior
CSI driver should attempt to use the block protocol that was specified for that PowerStore array.
Bug Description
Having defined two PowerStore arrays in the secret with different blockProtocol, the wrong protocol is selected during PVC creation. When the first array is set to NVMeTCP and the second to ISCSI the pvc, even when the storage class explicitly refers to the first one, attempt to connect to the volume via iSCSI which is not working since only NVMe host definitions have been created on the array. None of the appliances is set as "default".
Logs
{"level":"info","msg":"/csi.v1.Node/NodeStageVolume: REP 0009: rpc error: code = Internal desc = unable to find device after multiple discovery attempts: discovery complete but devices not found","time":"2024-09-25T11:54:28.889578221Z"}
Screenshots
No response
Additional Environment Information
No response
Steps to Reproduce
Environment preparation
Verify host definitions have been created with specified protocol on both the arrays.
Create workload which uses NVMeTCP storage class oc create -f workload_nvmetcp.yaml
Verify pods, pvc and pv are created oc get pod,pvc,pv -n
Result: pods are not being created as the CSI driver attempts to use iSCSI instead of NVMeTCP
Expected Behavior
CSI driver should attempt to use the block protocol that was specified for that PowerStore array.
CSM Driver(s)
csi-powerstore v2.11.1
Installation Type
No response
Container Storage Modules Enabled
None
Container Orchestrator
OpenShift 4.15.25
Operating System
Unknown