* Handled internally by a cloud provider as part of their services
* Others...

I'll give you more concrete examples in the next chapters.

## Restarts { #restarts }

Similar to making sure your application is run on startup, you probably also want to make sure it is **restarted** after failures.

### We Make Mistakes { #we-make-mistakes }

## Running on Startup and Restarts { #running-on-startup-and-restarts }

There is normally another tool in charge of **starting and running** your container.

It could be **Docker** directly, **Docker Compose**, **Kubernetes**, a **cloud service**, etc.

## Dynamic systems without restarting server

The following sub-systems are dynamic i.e., configuration parameters for each sub-systems can be changed while the server is running without any restarts.

```
api                   manage global HTTP API call specific features, such as throttling, authentication types, etc.
heal                  manage object healing frequency and bitrot verification checks

  * Automatically install updates on boot & reboot
  * Use optimized image based on Jessie by default
  * Add support for Ubuntu Wily
  * Master is configured with automatic restart-on-failure, via CloudWatch
  * Bootstrap reworked to be more similar to GCE; better supports reboots/restarts
  * Use an elastic IP for the master by default
  * Experimental support for node spot instances (set NODE\_SPOT\_PRICE=0.05)

#### GCE

// Manager will contain all the connections to the grid.
// It also handles incoming requests and routes them to the appropriate connection.
type Manager struct {
	// ID is an instance ID, that will change whenever the server restarts.
	// This allows remotes to keep track of whether state is preserved.
	ID uuid.UUID

	// Immutable after creation, so no locks.
	targets map[string]*Connection

	// serverside handlers.
	handlers handlers

			return nil, err
		}

		client := kes.NewClientWithConfig("", conf)
		client.Endpoints = endpoints

		// Keep the default key in the KES cache to prevent availability issues
		// when MinIO restarts
		go func() {
			timer := time.NewTicker(10 * time.Second)
			defer timer.Stop()
			defaultKey := env.Get(EnvKESDefaultKey, "")
			for {
				select {
				case <-ctx.Done():
					return

	}

	resourceID := "idmp-"
	roleID := env.Get(EnvIdentityPluginRoleID, kv.Get(RoleID))
	if roleID == "" {
		// We use a hash of the plugin URL so that the ARN remains
		// constant across restarts.
		h := sha1.New()
		h.Write([]byte(pluginURL))
		bs := h.Sum(nil)
		resourceID += base64.RawURLEncoding.EncodeToString(bs)
	} else {
		// Check that the roleID is restricted to URL safe characters

additionalAnnotations: {}

## Typically the deployment/statefulset includes checksums of secrets/config,
## So that when these change on a subsequent helm install, the deployment/statefulset
## is restarted. This can result in unnecessary restarts under GitOps tooling such as
## flux, so set to "true" to disable this behaviour.
ignoreChartChecksums: false

## Additional arguments to pass to minio binary
extraArgs: []

			Optional:    true,
			Type:        "on|off",
		},
		config.HelpKV{
			Key:         target.AmqpDurable,
			Description: "persist queue across broker restarts when set to 'on', default is 'off'",
			Optional:    true,
			Type:        "on|off",
		},
		config.HelpKV{
			Key:         target.AmqpNoWait,

	ResumeItemsFailed  uint64 `json:"-"`
	ResumeItemsSkipped uint64 `json:"-"`
	ResumeBytesDone    uint64 `json:"-"`
	ResumeBytesFailed  uint64 `json:"-"`
	ResumeBytesSkipped uint64 `json:"-"`

	// Filled on startup/restarts.
	QueuedBuckets []string

	// Filled during heal.
	HealedBuckets []string

	// ID of the current healing operation
	HealID string

	ItemsSkipped uint64
	BytesSkipped uint64

	RetryAttempts uint64