Description of problem:

Thanks to the work included on https://bugzilla.redhat.com/show_bug.cgi?id=2065597, now it is possible to configure cinder-csi by editing cloud-provider-config.

One of the parameters included is 'node-volume-attach-limit' [1]. Changing the param to 10 and running 11 pods connected to 11 different PVCs in the same worker is accepted:

$ oc get cm -n openshift-cluster-csi-drivers cloud-conf -o yaml
apiVersion: v1
data:
cloud.conf: |+
[Global]
region = regionOne
ca-file = /etc/kubernetes/static-pod-resources/configmaps/cloud-config/ca-bundle.pem
use-clouds = true
clouds-file = /etc/kubernetes/secret/clouds.yaml
cloud = openstack

[LoadBalancer]
use-octavia = True

[BlockStorage]
ignore-volume-az = yes
node-volume-attach-limit = 10

[Metadata]
search-order = metadataService

$ oc get pods
NAME READY STATUS RESTARTS AGE
demo-0-8466fc476-mbll5 1/1 Running 0 2m17s
demo-1-649948d68-dnhsv 1/1 Running 0 2m14s
demo-10-676c554976-52bwt 1/1 Running 0 103s
demo-2-6947d8d46d-mtxn6 1/1 Running 0 2m11s
demo-3-79db5fd4d8-qxgbl 1/1 Running 0 2m7s
demo-4-65678d5484-qlngk 1/1 Running 0 2m4s
demo-5-f75648d76-wzcmw 1/1 Running 0 2m
demo-6-856dc89dd8-s9bp4 1/1 Running 0 117s
demo-7-64f5944f8d-shq4d 1/1 Running 0 114s
demo-8-7b677bb644-4bhnn 1/1 Running 0 111s
demo-9-9b88b4c56-m7swp 1/1 Running 0 107s
$ oc get pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
pvc-0 Bound pvc-92aaeab8-6c60-4f32-9ab6-a703761ad3cd 1Gi RWO topology-aware-0 2m19s
pvc-1 Bound pvc-336915af-8154-4e67-b37d-2fa93f2db645 1Gi RWO topology-aware-1 2m16s
pvc-10 Bound pvc-a7f58825-e465-4e2b-b9ec-36362589d1d2 1Gi RWO topology-aware-10 105s
pvc-2 Bound pvc-4ded3dab-170d-4573-ba29-8bb10b3e857b 1Gi RWO topology-aware-2 2m13s
pvc-3 Bound pvc-b3d7ec16-5b7e-4964-9a0a-29480dee31a1 1Gi RWO topology-aware-3 2m9s
pvc-4 Bound pvc-a6ae7415-00b5-4849-8065-7fe7acace165 1Gi RWO topology-aware-4 2m6s
pvc-5 Bound pvc-246421b3-2716-41c8-88e9-d6253c49acdd 1Gi RWO topology-aware-5 2m2s
pvc-6 Bound pvc-61ec111a-d5ae-47d2-83be-1a4a344afea7 1Gi RWO topology-aware-6 119s
pvc-7 Bound pvc-b65cd362-977a-4613-84a3-bb75b1adf971 1Gi RWO topology-aware-7 116s
pvc-8 Bound pvc-a79d7bb1-9697-4c1b-8ad3-696a0c2edafd 1Gi RWO topology-aware-8 113s
pvc-9 Bound pvc-48e37bc7-ff50-4517-8cac-a57fd6f0f47b 1Gi RWO topology-aware-9 109s
$ openstack volume list -c 'Attached to' | grep ostest | awk -F' ' '

' | sort | uniq -c
11 ostest-zh8g7-worker-0-7gl4r
$ openstack volume list
--------------------------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------------------------------------------------------------------------------------

[1] https://github.com/kubernetes/cloud-provider-openstack/blob/master/docs/cinder-csi-plugin/using-cinder-csi-plugin.md#driver-config

Version-Release number of selected component (if applicable): 4.11.0-0.nightly-2022-06-06-025509 RHOS-16.2-RHEL-8-20220311.n.1

How reproducible: Always

Steps to Reproduce:
1. Change cloud cm to node-volume-attach-limit = 10
2. Wait until cluster re-configuration is completed (around 30 minutes)
3. Create 11 pods using cinder PVCs with NodeSelector set to use always the same worker.

Actual results: The node-volume-attach-limit is ignored.
Expected results: The PVC number 11 should not becreated.