Get desktop application:
View/edit binary Protocol Buffers messages
Tell the object store to deallocate an object held by the object store. This is called by the scheduler.
The canonical objectid of the object to deallocate
Notify the object store about objectid aliasing. This is called by the scheduler
The objectid being aliased
The canonical objectid that points to the actual object
Get info about the object store state
Object IDs we want to retrieve from the store for inspection
List of object IDs in the store
Protocol buffer objects that were requested
Tell the object store to begin getting an object from another object store (called by the scheduler)
Object store to get the object from
ID of object that gets delivered
Accept incoming data from another object store, as a stream of object chunks
Object ID of the object being streamed
Total size of the object
Offset of the arrow metadata
Data for this chunk of the object
Used by the worker to notify the scheduler about which objectids a particular object contains
The objectid of the object in question
Object IDs contained in the object
Used by the worker to tell the scheduler that two objectids should refer to the same object
ObjectID which will be aliased
The target ObjectID
Decrement the count of the object ID
Decrements the reference count of a particular object ID
Object IDs whose reference count should be decremented. Duplicates will be decremented multiple times.
Exports function to the workers
Ship an initializer and reinitializer for a reusable variable to the workers
The reusable variable to export.
Increment the count of the object ID
Increments the reference count of a particular object ID
Object IDs whose reference count should be incremented. Duplicates will be incremented multiple times.
Kills the workers
(message has no fields)
Currently, the only reason to fail is if there are workers still executing tasks
Notify the scheduler that a failure occurred while running a task, importing a remote function, or importing a reusable variable.
The failure object.
Used by an object store to tell the scheduler that an object is ready (i.e. has been finalized and can be shared)
Object ID of the object that has been finalized
ID of the object store the object lives on
Request an object ID for an object that will be put in an object store
Worker that tries to put an object
Object ID assigned by the scheduler to the object
Used by the worker to ask for work, this also returns the status of the previous task if there was one
ID of the worker which executed the task
Register an object store with the scheduler
IP address of the object store being registered
Object store ID assigned by the scheduler
Tell the scheduler that a worker successfully imported a remote function.
Worker that can execute the function
Name of the remote function
Number of return values of the function. This is only present if the function was successfully imported.
Register a new worker with the scheduler
The IP address of the node the worker is running on.
The address of the worker.
The address of the object store the worker should connect to. If omitted, this will be assigned by the scheduler.
True if the worker is a driver, and false otherwise.
Worker ID assigned by the scheduler
The Object store ID of the worker's local object store
IP address of the object store the worker should connect to
Request delivery of an object from an object store that holds the object to the local object store
Worker that tries to request the object
Object ID of the object being requested
Run a function on all workers
Get information about the scheduler state
(message has no fields)
OperationIds of the tasks on the task queue
List of workers waiting to get a task assigned
Table of all available remote function
The target_objectids_ data structure
The reference_counts_ data structure
The computation graph constructed so far
Information about the object stores
Asks the scheduler to execute a task, immediately returns an object ID to the result
The ID of the worker submitting the task
Contains name of the function to be executed and arguments
Object IDs of the function return values
True if the function was registered, false otherwise
True if no workers have registered with the scheduler, false otherwise
Get information about tasks
(message has no fields)
The tasks that have failed.
The tasks that are currently running.
The remote function imports that failed.
The reusable variable imports that failed.
The reusable variable reinitializations that failed.
The function to run on all workers that failed.
Polls the scheduler to see what objectids can be retrieved in the input list.
List of objectids to be checked.
List of indices that correspond to objectids in the original list that are ready.
Kills this worker
(message has no fields)
Scheduler calls a function from the worker
Contains name of the function to be executed and arguments
Scheduler imports a function into the worker
Scheduler imports a reusable variable into the worker
The reusable variable to export.
Causes an error message to be printed.
The failure object.
Runs a function on the worker.
Used as response type in: ObjStore.DeallocateObject, ObjStore.NotifyAlias, ObjStore.StartDelivery, Scheduler.AddContainedObjectIDs, Scheduler.AliasObjectIDs, Scheduler.DecrementCount, Scheduler.DecrementRefCount, Scheduler.ExportRemoteFunction, Scheduler.ExportReusableVariable, Scheduler.IncrementCount, Scheduler.IncrementRefCount, Scheduler.NotifyFailure, Scheduler.ObjReady, Scheduler.ReadyForNewTask, Scheduler.RegisterRemoteFunction, Scheduler.RequestObj, Scheduler.RunFunctionOnAllWorkers, WorkerService.Die, WorkerService.ExecuteTask, WorkerService.ImportRemoteFunction, WorkerService.ImportReusableVariable, WorkerService.PrintErrorMessage, WorkerService.RunFunctionOnWorker
(message has no fields)
Used in:
The objectid for the argument.
A serialized representation of an argument passed by value.
Used in:
Used in:
Used as request type in: Scheduler.DecrementCount, Scheduler.IncrementCount
Object ID of the object whose reference count is increased or decreased
Used in:
Used in:
Used in:
Used in:
Empty used to represent a None object
Used in:
(message has no fields)
Used in:
Used to represent exceptions thrown in Python. This will happen when a task fails to execute, a remote function fails to be imported, or a reusable variable fails to be imported.
Used in: , , ,
The type of the failure.
The id of the worker on which the failure occurred.
The address of the worker on which the failure occurred. This contains the same information as the workerid.
The name of the failed object.
The error message from the failure.
Used in:
ID of the worker that can execute the function
Number of return values of the function
Used for shipping remote functions to workers
Used in: , , , , ,
Used in:
The objectid for the object that is retrieved
Object ID of the object being requested by the worker
Used in:
Used in:
Used in:
The failure object.
Union of possible object types
Used in: , , ,
Used in:
Used in:
The ID of the object store.
The address of the object store.
This is used internally by the scheduler. From the scheduler's perspective, the submission of tasks (via SubmitTask) and the submission of puts (via PutObj) look very similar, and so it is useful to be able to handle them together (for example in the computation graph).
Used in:
The id of the task that called this task or put.
Used in:
The objectid for the object that was put
Used in:
Handle to memory segment where object is stored
ID of object that gets registered
Used in: , ,
The name of the reusable variable.
A serialized version of the function that initializes the reusable variable.
A serialized version of the function that reinitializes the reusable variable.
Used in:
Used in: , , ,
Name of the function call. Must not be empty.
List of object IDs of the arguments to the function.
Object IDs for result
Used in:
Used in:
Used in:
This message is used by the worker service to send messages to the worker that are processed by the worker's main loop.
A task for the worker to execute.
A remote function to import on the worker.
A reusable variable to import on the worker.
An arbitrary function to run on the worker.