buildgrid.server.execution.service module

class buildgrid.server.execution.service.ExecutionService

Bases: ExecutionServicer, InstancedServicer[ExecutionInstance]

SERVICE_NAME = 'Execution'
REGISTER_METHOD(server)
FULL_NAME: ClassVar[str] = 'build.bazel.remote.execution.v2.Execution'

The full name of the servicer, used to match instances to the servicer and configure reflection. This value should be declared on the class of any Servicer implementations.

Execute(request: ExecuteRequest, context: ServicerContext) Iterator[Operation]

Execute an action remotely.

In order to execute an action, the client must first upload all of the inputs, the [Command][build.bazel.remote.execution.v2.Command] to run, and the [Action][build.bazel.remote.execution.v2.Action] into the [ContentAddressableStorage][build.bazel.remote.execution.v2.ContentAddressableStorage]. It then calls Execute with an action_digest referring to them. The server will run the action and eventually return the result.

The input Action’s fields MUST meet the various canonicalization requirements specified in the documentation for their types so that it has the same digest as other logically equivalent Action`s. The server MAY enforce the requirements and return errors if a non-canonical input is received. It MAY also proceed without verifying some or all of the requirements, such as for performance reasons. If the server does not verify the requirement, then it will treat the `Action as distinct from another logically equivalent action if they hash differently.

Returns a stream of [google.longrunning.Operation][google.longrunning.Operation] messages describing the resulting execution, with eventual response [ExecuteResponse][build.bazel.remote.execution.v2.ExecuteResponse]. The metadata on the operation is of type [ExecuteOperationMetadata][build.bazel.remote.execution.v2.ExecuteOperationMetadata].

If the client remains connected after the first response is returned after the server, then updates are streamed as if the client had called [WaitExecution][build.bazel.remote.execution.v2.Execution.WaitExecution] until the execution completes or the request reaches an error. The operation can also be queried using [Operations API][google.longrunning.Operations.GetOperation].

The server NEED NOT implement other methods or functionality of the Operations API.

Errors discovered during creation of the Operation will be reported as gRPC Status errors, while errors that occurred while running the action will be reported in the status field of the ExecuteResponse. The server MUST NOT set the error field of the Operation proto. The possible errors include:

  • INVALID_ARGUMENT: One or more arguments are invalid.

  • FAILED_PRECONDITION: One or more errors occurred in setting up the

action requested, such as a missing input or command or no worker being available. The client may be able to fix the errors and retry. * RESOURCE_EXHAUSTED: There is insufficient quota of some resource to run the action. * UNAVAILABLE: Due to a transient condition, such as all workers being occupied (and the server does not support a queue), the action could not be started. The client should retry. * INTERNAL: An internal error occurred in the execution engine or the worker. * DEADLINE_EXCEEDED: The execution timed out. * CANCELLED: The operation was cancelled by the client. This status is only possible if the server implements the Operations API CancelOperation method, and it was called for the current execution.

In the case of a missing input or command, the server SHOULD additionally send a [PreconditionFailure][google.rpc.PreconditionFailure] error detail where, for each requested blob not present in the CAS, there is a Violation with a type of MISSING and a subject of “blobs/{digest_function/}{hash}/{size}” indicating the digest of the missing blob. The subject is formatted the same way as the resource_name provided to [ByteStream.Read][google.bytestream.ByteStream.Read], with the leading instance name omitted. digest_function MUST thus be omitted if its value is one of MD5, MURMUR3, SHA1, SHA256, SHA384, SHA512, or VSO.

The server does not need to guarantee that a call to this method leads to at most one execution of the action. The server MAY execute the action multiple times, potentially in parallel. These redundant executions MAY continue to run, even if the operation is completed.

WaitExecution(request: WaitExecutionRequest, context: ServicerContext) Iterator[Operation]

Wait for an execution operation to complete. When the client initially makes the request, the server immediately responds with the current status of the execution. The server will leave the request stream open until the operation completes, and then respond with the completed operation. The server MAY choose to stream additional updates as execution progresses, such as to provide an update as to the state of the execution.

In addition to the cases describe for Execute, the WaitExecution method may fail as follows:

  • NOT_FOUND: The operation no longer exists due to any of a transient

condition, an unknown operation name, or if the server implements the Operations API DeleteOperation method and it was called for the current execution. The client should call Execute to retry.

query_connected_clients_for_instance(instance_name: str) int