When setting the following for "collection" in "ClusterLogging", the Cluster Logging Operator crashes:

~~~
apiVersion: logging.openshift.io/v1
kind: ClusterLogging
metadata:
  annotations:
    logging.openshift.io/preview-vector-collector: enabled
  name: instance
  namespace: openshift-logging
spec:
[..]
  collection: {}
[..]
~~~

The Operator then crashes with the following stack trace:

~~~
{"_ts":"2023-03-03T12:38:30.51672121Z","_level":"0","_component":"cluster-logging-operator","_message":"starting up...","go_arch":"amd64","go_os":"linux","go_version":"go1.18.7","operator_version":"5.6"}
I0303 12:38:31.568169 1 request.go:665] Waited for 1.038650902s due to client-side throttling, not priority and fairness, request: GET:https://172.30.0.1:443/apis/monitoring.coreos.com/v1?timeout=32s
{"_ts":"2023-03-03T12:38:33.476748674Z","_level":"0","_component":"cluster-logging-operator","_message":"migrating resources provided by the manifest"}
{"_ts":"2023-03-03T12:38:33.48144877Z","_level":"0","_component":"cluster-logging-operator","_message":"Registering Components."}
{"_ts":"2023-03-03T12:38:33.48156936Z","_level":"0","_component":"cluster-logging-operator","_message":"Starting the Cmd."}
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x1426670]
goroutine 580 [running]:
github.com/openshift/cluster-logging-operator/internal/migrations.MigrateCollectionSpec({{0xc000592610, 0x7}, 0xc0003a05f0, 0xc00026d7d0, 0xc0003a0640, 0x0, 0x0})
/remote-source/cluster-logging-operator/app/internal/migrations/migrate_collection.go:28 +0x2d0
github.com/openshift/cluster-logging-operator/internal/k8shandler.(*ClusterLoggingRequest).getClusterLogging(0xc000952300, 0x0)
/remote-source/cluster-logging-operator/app/internal/k8shandler/reconciler.go:218 +0x17d
github.com/openshift/cluster-logging-operator/internal/k8shandler.Reconcile(0xc0004285a0, {0x19fa270, 0xc0000bec30}, {0x7fd9b0687628, 0xc00017a1c0}, {0x19f47e8, 0xc0002e0040}, {0xc000432e99, 0x7}, {0xc000628690, ...})
/remote-source/cluster-logging-operator/app/internal/k8shandler/reconciler.go:47 +0x20b
github.com/openshift/cluster-logging-operator/controllers/clusterlogging.(*ReconcileClusterLogging).Reconcile(0xc0002bb200, {0x19f5348, 0xc00026c0f0}, {{{0xc00038dce0?, 0x16b4fe0?}, {0xc0005925d0?, 0x30?}}})
/remote-source/cluster-logging-operator/app/controllers/clusterlogging/clusterlogging_controller.go:88 +0x3c9
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).Reconcile(0xc00077a000, {0x19f5348, 0xc00026c0c0}, {{{0xc00038dce0?, 0x16b4fe0?}, {0xc0005925d0?, 0x409514?}}})
/remote-source/cluster-logging-operator/deps/gomod/pkg/mod/sigs.k8s.io/controller-runtime@v0.11.0/pkg/internal/controller/controller.go:114 +0x27e
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).reconcileHandler(0xc00077a000, {0x19f52a0, 0xc0000ba280}, {0x15f4900?, 0xc0006bc040?})
/remote-source/cluster-logging-operator/deps/gomod/pkg/mod/sigs.k8s.io/controller-runtime@v0.11.0/pkg/internal/controller/controller.go:311 +0x349
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).processNextWorkItem(0xc00077a000, {0x19f52a0, 0xc0000ba280})
/remote-source/cluster-logging-operator/deps/gomod/pkg/mod/sigs.k8s.io/controller-runtime@v0.11.0/pkg/internal/controller/controller.go:266 +0x1d9
sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).Start.func2.2()
/remote-source/cluster-logging-operator/deps/gomod/pkg/mod/sigs.k8s.io/controller-runtime@v0.11.0/pkg/internal/controller/controller.go:227 +0x85
created by sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).Start.func2
/remote-source/cluster-logging-operator/deps/gomod/pkg/mod/sigs.k8s.io/controller-runtime@v0.11.0/pkg/internal/controller/controller.go:223 +0x31c
~~~

* OpenShift Container Platform 4.11.21
* cluster-logging.v5.6.2

Always

1. Deploy Cluster Logging 5.6.2
2. Configure the ClusterLogging CRD with a proper configuration, use the ClusterLogging definition from the documentation for example
3. After the stack has been deployed successfully, change the "collection" field to "{}"

Operator crashes with "panic: runtime error: invalid memory address or nil pointer dereference"

Operator ignores the invalid configuration or warns about it

* Customer noticed this when trying out Vector and wanted to remove his fluentd Pods first. Workaround is to completely remove the "collection" field, then the Pods are removed as expected.