WIP: dump cluster logs

cmd/hostpathplugin/main.go

@@ -52,7 +52,6 @@ func main() {
 	flag.Int64Var(&cfg.MaxVolumesPerNode, "maxvolumespernode", 0, "limit of volumes per node")
 	flag.Var(&cfg.Capacity, "capacity", "Simulate storage capacity. The parameter is <kind>=<quantity> where <kind> is the value of a 'kind' storage class parameter and <quantity> is the total amount of bytes for that kind. The flag may be used multiple times to configure different kinds.")
 	flag.BoolVar(&cfg.EnableAttach, "enable-attach", false, "Enables RPC_PUBLISH_UNPUBLISH_VOLUME capability.")
-	flag.BoolVar(&cfg.CheckVolumeLifecycle, "check-volume-lifecycle", false, "Can be used to turn some violations of the volume lifecycle into warnings instead of failing the incorrect gRPC call. Disabled by default because of https://github.com/kubernetes/kubernetes/issues/101911.")
 	flag.Int64Var(&cfg.MaxVolumeSize, "max-volume-size", 1024*1024*1024*1024, "maximum size of volumes in bytes (inclusive)")
 	flag.BoolVar(&cfg.EnableTopology, "enable-topology", true, "Enables PluginCapability_Service_VOLUME_ACCESSIBILITY_CONSTRAINTS capability.")
 	flag.BoolVar(&cfg.EnableVolumeExpansion, "node-expand-required", true, "Enables NodeServiceCapability_RPC_EXPAND_VOLUME capacity.")

go.mod

@@ -15,7 +15,6 @@ require (
 	google.golang.org/genproto v0.0.0-20201209185603-f92720507ed4 // indirect
 	google.golang.org/grpc v1.34.0
 	k8s.io/apimachinery v0.21.0-alpha.0
-	k8s.io/klog/v2 v2.4.0
 	k8s.io/kubernetes v1.20.0
 	k8s.io/mount-utils v0.20.0 // indirect
 	k8s.io/utils v0.0.0-20201110183641-67b214c5f920

pkg/hostpath/controllerserver.go

@@ -33,7 +33,6 @@ import (
 	"google.golang.org/grpc/status"
 
 	"github.com/container-storage-interface/spec/lib/go/csi"
-	"k8s.io/klog/v2"
 	utilexec "k8s.io/utils/exec"
 
 	"github.com/kubernetes-csi/csi-driver-host-path/pkg/state"
@@ -205,12 +204,8 @@ func (hp *hostPath) DeleteVolume(ctx context.Context, req *csi.DeleteVolumeReque
 	}
 
 	if vol.Attached || !vol.Published.Empty() || !vol.Staged.Empty() {
-		msg := fmt.Sprintf("Volume '%s' is still used (attached: %v, staged: %v, published: %v) by '%s' node",
+		return nil, status.Errorf(codes.Internal, "Volume '%s' is still used (attached: %v, staged: %v, published: %v) by '%s' node",
 			vol.VolID, vol.Attached, vol.Staged, vol.Published, vol.NodeID)
-		if hp.config.CheckVolumeLifecycle {
-			return nil, status.Error(codes.Internal, msg)
-		}
-		klog.Warning(msg)
 	}
 
 	if err := hp.deleteVolume(volId); err != nil {
@@ -345,12 +340,8 @@ func (hp *hostPath) ControllerUnpublishVolume(ctx context.Context, req *csi.Cont
 
 	// Check to see if the volume is staged/published on a node
 	if !vol.Published.Empty() || !vol.Staged.Empty() {
-		msg := fmt.Sprintf("Volume '%s' is still used (staged: %v, published: %v) by '%s' node",
+		return nil, status.Errorf(codes.Internal, "Volume '%s' is still used (staged: %v, published: %v) by '%s' node",
 			vol.VolID, vol.Staged, vol.Published, vol.NodeID)
-		if hp.config.CheckVolumeLifecycle {
-			return nil, status.Error(codes.Internal, msg)
-		}
-		klog.Warning(msg)
 	}
 
 	vol.Attached = false

pkg/hostpath/hostpath.go

@@ -74,7 +74,6 @@ type Config struct {
 	EnableAttach          bool
 	EnableTopology        bool
 	EnableVolumeExpansion bool
-	CheckVolumeLifecycle  bool
 }
 
 var (

release-tools/prow.sh

@@ -609,7 +609,21 @@ apiVersion: kind.x-k8s.io/v1alpha4
 nodes:
 - role: control-plane
 - role: worker
+  kubeadmConfigPatches:
+  - |
+    kind: JoinConfiguration
+    nodeRegistration:
+      kubeletExtraArgs:
+        v: "5"
+        vmodule: "*csi*=5"
 - role: worker
+  kubeadmConfigPatches:
+  - |
+    kind: JoinConfiguration
+    nodeRegistration:
+      kubeletExtraArgs:
+        v: "5"
+        vmodule: "*csi*=5"
 featureGates:
 $(list_gates "$gates")
 runtimeConfig:
@@ -632,11 +646,16 @@ EOF
 
 # Deletes kind cluster inside a prow job
 delete_cluster_inside_prow_job() {
+    local name="$1"
+
     # Inside a real Prow job it is better to clean up at runtime
     # instead of leaving that to the Prow job cleanup code
     # because the later sometimes times out (https://github.com/kubernetes-csi/csi-release-tools/issues/24#issuecomment-554765872).
+    #
+    # This is also a good time to collect logs.
     if [ "$JOB_NAME" ]; then
         if kind get clusters | grep -q csi-prow; then
+            run kind export logs --name=csi-prow ${ARTIFACTS}/cluster-logs/$name
             run kind delete cluster --name=csi-prow || die "kind delete failed"
         fi
         unset KUBECONFIG
@@ -1252,7 +1271,7 @@ main () {
                     fi
                 fi
             fi
-            delete_cluster_inside_prow_job
+            delete_cluster_inside_prow_job non-alpha
         fi
 
         if tests_need_alpha_cluster && [ "${CSI_PROW_E2E_ALPHA_GATES}" ]; then
@@ -1287,7 +1306,7 @@ main () {
                     fi
                 fi
             fi
-            delete_cluster_inside_prow_job
+            delete_cluster_inside_prow_job alpha
         fi
     fi
 

vendor/modules.txt

@@ -154,7 +154,6 @@ k8s.io/apiserver/pkg/util/feature
 # k8s.io/component-base v0.20.0 => k8s.io/component-base v0.20.0
 k8s.io/component-base/featuregate
 # k8s.io/klog/v2 v2.4.0
-## explicit
 k8s.io/klog/v2
 # k8s.io/kubernetes v1.20.0
 ## explicit