{{- else if .Values.cni.repair.deletePods }}
  - apiGroups: [""]
    resources: ["pods"]
    verbs: ["delete"]
{{- else if .Values.cni.repair.labelPods }}
  - apiGroups: [""]
    {{- /* pods/status is less privileged than the full pod, and either can label. So use the lower pods/status */}}
    resources: ["pods/status"]
    verbs: ["patch", "update"]
{{- end }}
{{- end }}
---
{{- if .Values.cni.ambient.enabled }}

under the `kubelet` process. On systems with `journalctl` the following is an example command line
to view the last 1000 `kubelet` logs via the `less` utility to allow for `vi`-style searching:

```console
$ journalctl -t kubelet -n 1000 | less
```

#### GKE via Stackdriver Log Viewer

Each GKE cluster's will have many categories of logs collected by Stackdriver.  Logs can be monitored via

}

// ObjectTooSmall error returned when the size of the object < what is expected.
type ObjectTooSmall GenericError

func (e ObjectTooSmall) Error() string {
	return "size of the object less than what is expected"
}

// OperationTimedOut - a timeout occurred.
type OperationTimedOut struct{}

func (e OperationTimedOut) Error() string {
	return "Operation timed out"
}

When you run `fastapi dev`, it will run on development mode.

By default, it will have **auto-reload** enabled, so it will automatically reload the server when you make changes to your code. This is resource intensive and could be less stable than without it, you should only use it for development.

By default it will listen on the IP address `127.0.0.1`, which is the IP for your machine to communicate with itself alone (`localhost`).

## `fastapi run`

* **Intuitive**: Great editor support. <abbr title="also known as auto-complete, autocompletion, IntelliSense">Completion</abbr> everywhere. Less time debugging.
* **Easy**: Designed to be easy to use and learn. Less time reading docs.
* **Short**: Minimize code duplication. Multiple features from each parameter declaration. Fewer bugs.

// sort.Sort call, can be used to prepare arbitrary inputs for use as sets.
func uniq(data sort.Interface) (size int) {
	p, l := 0, data.Len()
	if l <= 1 {
		return l
	}
	for i := 1; i < l; i++ {
		if !data.Less(p, i) {
			continue
		}
		p++
		if p < i {
			data.Swap(p, i)
		}
	}
	return p + 1
}

// Remove any duplicates and return unique entries.
func uniqueEntries(paths []string) []string {

* **Intuitive**: Great editor support. <abbr title="also known as auto-complete, autocompletion, IntelliSense">Completion</abbr> everywhere. Less time debugging.
* **Easy**: Designed to be easy to use and learn. Less time reading docs.
* **Short**: Minimize code duplication. Multiple features from each parameter declaration. Fewer bugs.

* Running on startup
* Restarts
* Replication (the number of processes running)
* Memory
* Previous steps before starting

## Memory

If you run **a single process per container** you will have a more or less well-defined, stable, and limited amount of memory consumed by each of those containers (more than one if they are replicated).

	dataUsageUpdateDirCycles         = 16                               // Visit all folders every n cycles.
	dataScannerCompactLeastObject    = 500                              // Compact when there is less than this many objects in a branch.
	dataScannerCompactAtChildren     = 10000                            // Compact when there are this many children in a branch.

An interesting detail is that the percentage of the **CPU used** by each process can **vary** a lot over time, but the **memory (RAM)** normally stays more or less **stable**.

If you have an API that does a comparable amount of computations every time and you have a lot of clients, then the **CPU utilization** will probably *also be stable* (instead of constantly going up and down quickly).