// can have their pods stopped for an update at any given time. The update
  // starts by stopping at most 30% of those DaemonSet pods and then brings
  // up new DaemonSet pods in their place. Once the new pods are available,
  // it then proceeds onto other DaemonSet pods, thus ensuring that at least
  // 70% of original number of DaemonSet pods are available at all times during
  // the update.

  // can have their pods stopped for an update at any given time. The update
  // starts by stopping at most 30% of those DaemonSet pods and then brings
  // up new DaemonSet pods in their place. Once the new pods are available,
  // it then proceeds onto other DaemonSet pods, thus ensuring that at least
  // 70% of original number of DaemonSet pods are available at all times during
  // the update.

  optional k8s.io.apimachinery.pkg.apis.meta.v1.GroupVersionKind kind = 2;

  // Resource is the fully-qualified resource being requested (for example, v1.pods)
  optional k8s.io.apimachinery.pkg.apis.meta.v1.GroupVersionResource resource = 3;

  // SubResource is the subresource being requested, if any (for example, "status" or "scale")
  // +optional

  // can have their pods stopped for an update at any given time. The update
  // starts by stopping at most 30% of those DaemonSet pods and then brings
  // up new DaemonSet pods in their place. Once the new pods are available,
  // it then proceeds onto other DaemonSet pods, thus ensuring that at least
  // 70% of original number of DaemonSet pods are available at all times during
  // the update.

1. By running as a node-level daemonset that:

- listens for these UDS events from the CNI plugin (which fire when new pods are spawned in an ambient-enabled namespace), and adds those pods to the ambient mesh.
- watches k8s resource for existing pods, so that pods that have already been started can be moved in or out of the ambient mesh.

  // +optional
  optional k8s.io.apimachinery.pkg.apis.meta.v1.ObjectMeta metadata = 1;

  // handler specifies the underlying runtime and configuration that the CRI
  // implementation will use to handle pods of this class. The possible values
  // are specific to the node & CRI configuration.  It is assumed that all
  // handlers are available on every node, and handlers of the same name are
  // equivalent on every node.

  // value represents the integer value of this priority class. This is the actual priority that pods
  // receive when they have the name of this class in their pod spec.
  optional int32 value = 2;

  // globalDefault specifies whether this PriorityClass should be considered as
  // the default priority for pods that do not have any priority class.
  // Only one PriorityClass can be marked as `globalDefault`. However, if more than

  optional k8s.io.apimachinery.pkg.apis.meta.v1.GroupVersionKind kind = 2;

  // Resource is the fully-qualified resource being requested (for example, v1.pods)
  optional k8s.io.apimachinery.pkg.apis.meta.v1.GroupVersionResource resource = 3;

  // SubResource is the subresource being requested, if any (for example, "status" or "scale")
  // +optional

如果你有 **多个容器**，可能每个容器都运行一个 **单个进程**（例如，在 **Kubernetes** 集群中），那么你可能希望有一个 **单独的容器** 执行以下操作： 在单个容器中运行单个进程执行**先前步骤**，即运行复制的worker容器之前。

!!! info
    如果你使用 Kubernetes，这可能是 <a href="https://kubernetes.io/docs/concepts/workloads/pods/init-containers/" class="external-link" target="_blank">Init Container</a>。

如果在你的用例中，运行前面的步骤**并行多次**没有问题（例如，如果你没有运行数据库迁移，而只是检查数据库是否已准备好），那么你也可以将它们放在开始主进程之前在每个容器中。

### 单容器

        multiple of the number of replicas.
    *   `tf.data.Dataset` now supports automatic data distribution and sharding
        in distributed environments, including on TPU pods.
    *   Distribution policies for `tf.data.Dataset` can now be tuned with 1.
        `tf.data.experimental.AutoShardPolicy(OFF, AUTO, FILE, DATA)` 2.
        `tf.data.experimental.ExternalStatePolicy(WARN, IGNORE, FAIL)`