Upgrade tests failing in master from file test pod already existing

[travisn]

Is this a bug report or feature request?

• Bug Report

Deviation from expected behavior:
The upgrade tests are failing in the master branch since #8758 was merged. They must have a different permutation than what is run for PRs. For example, see this build.

Error from server (AlreadyExists): error when creating "STDIN": persistentvolumeclaims "file-test" already exists
Error from server (AlreadyExists): error when creating "STDIN": pods "file-test" already exists
2021-09-27 13:47:40.956120 E | utils: Failed to execute stdin: kubectl [create -f -] : exit status 1
2021-09-27 13:47:40.956149 E | testutil: Failed to execute kubectl [create -f -] -- Failed to run stdin: kubectl [create -f -] : Error from server (AlreadyExists): error when creating "STDIN": persistentvolumeclaims "file-test" already existsError from server (AlreadyExists): error when creating "STDIN": pods "file-test" already exists
    ceph_base_file_test.go:291: 
        	Error Trace:	ceph_base_file_test.go:291
        	            				ceph_upgrade_test.go:255
        	            				ceph_upgrade_test.go:102
        	Error:      	Received unexpected error:
        	            	failed to create pod -- 
        	            	apiVersion: v1
        	            	kind: PersistentVolumeClaim
        	            	metadata:
        	            	  name: file-test
        	            	  namespace: upgrade-ns
        	            	spec:
        	            	  accessModes:
        	            	  - ReadWriteOnce
        	            	  resources:
        	            	    requests:
        	            	      storage: 1Gi
        	            	  storageClassName: file-upgrade
        	            	---
        	            	apiVersion: v1
        	            	kind: Pod
        	            	metadata:
        	            	  name: file-test
        	            	  namespace: upgrade-ns
        	            	spec:
        	            	  containers:
        	            	  - name: file-test
        	            	    image: busybox
        	            	    command:
        	            	        - sh
        	            	        - "-c"
        	            	        - "touch /tmp/testrook/csi.test && sleep 3600"
        	            	    imagePullPolicy: IfNotPresent
        	            	    env:
        	            	    volumeMounts:
        	            	    - mountPath: /tmp/testrook
        	            	      name: csivol
        	            	  volumes:
        	            	  - name: csivol
        	            	    persistentVolumeClaim:
        	            	       claimName: file-test
        	            	       readOnly: false
        	            	  restartPolicy: Never
        	            	. Could Not create resource in args : [create -f -] -- Failed to run stdin: kubectl [create -f -] : Error from server (AlreadyExists): error when creating "STDIN": persistentvolumeclaims "file-test" already existsError from server (AlreadyExists): error when creating "STDIN": pods "file-test" already exists
        	Test:       	TestCephUpgradeSuite/TestUpgradeRookToMaster

Expected behavior:
Passing integration tests.

How to reproduce it (minimal and precise):

Trigger the CI in the master branch