'mlir_module' is a serialized MLIR module with a `main` function that contains
target computation.
'metadata' is a serialized TPUCompileMetadataProto describing the shapes and
types of the inputs to the computation, as well as a mapping onto the TPU pod
topology.
'producer_name' is a string describing the name of the framework that add support for running this portion of the model on TPUs.
  }];

  let arguments = (ins

  if (${env:CONTAINER_RUNTIME_ENDPOINT}) {
    & "${env:NODE_DIR}\crictl.exe" config runtime-endpoint `
        ${env:CONTAINER_RUNTIME_ENDPOINT}
  }
}

# Pulls the infra/pause container image onto the node so that it will be
# immediately available when the kubelet tries to run pods.
# TODO(pjh): downloading the container container image may take a few minutes;

[[sec:unpacking_archives_example]]
== Unpacking archives

Archives are effectively self-contained file systems, so unpacking them is a case of copying the files from that file system onto the local file system — or even into another archive.
Gradle enables this by providing some wrapper functions that make archives available as hierarchical collections of files (<<sec:file_trees,file trees>>).

### Action Required

* Add direct serializer ([#26251](https://github.com/kubernetes/kubernetes/pull/26251), [@caesarxuchao](https://github.com/caesarxuchao))
* Add a NodeCondition "NetworkUnavailable" to prevent scheduling onto a node until the routes have been created  ([#26415](https://github.com/kubernetes/kubernetes/pull/26415), [@wojtek-t](https://github.com/wojtek-t))

	// which should be the final remaining step needed for the link.
	// This will reduce peak RSS for the link (and speed up linking of
	// large applications), since when the archive command runs we
	// won't be holding onto all of the linker's live memory.
	if syscallExecSupported && !ownTmpDir {
		runAtExitFuncs()
		ctxt.execArchive(argv)
		panic("should not get here")
	}