Create a controller with an AWS Batch launcher.

Create an R6 object to submit tasks and launch workers on AWS Batch workers.

Usage

crew_controller_aws_batch(
  name = NULL,
  workers = 1L,
  host = NULL,
  port = NULL,
  tls = crew::crew_tls(mode = "automatic"),
  tls_enable = NULL,
  tls_config = NULL,
  serialization = NULL,
  profile = crew::crew_random_name(),
  seconds_interval = 0.5,
  seconds_timeout = 60,
  seconds_launch = 1800,
  seconds_idle = 300,
  seconds_wall = Inf,
  retry_tasks = NULL,
  tasks_max = Inf,
  tasks_timers = 0L,
  reset_globals = TRUE,
  reset_packages = FALSE,
  reset_options = FALSE,
  garbage_collection = FALSE,
  crashes_error = NULL,
  processes = NULL,
  r_arguments = c("--no-save", "--no-restore"),
  crashes_max = 5L,
  backup = NULL,
  options_metrics = crew::crew_options_metrics(),
  options_aws_batch = crew.aws.batch::crew_options_aws_batch(),
  aws_batch_config = NULL,
  aws_batch_credentials = NULL,
  aws_batch_endpoint = NULL,
  aws_batch_region = NULL,
  aws_batch_job_definition = NULL,
  aws_batch_job_queue = NULL,
  aws_batch_share_identifier = NULL,
  aws_batch_scheduling_priority_override = NULL,
  aws_batch_parameters = NULL,
  aws_batch_container_overrides = NULL,
  aws_batch_node_overrides = NULL,
  aws_batch_retry_strategy = NULL,
  aws_batch_propagate_tags = NULL,
  aws_batch_timeout = NULL,
  aws_batch_tags = NULL,
  aws_batch_eks_properties_override = NULL
)

Arguments

name

Character string, name of the launcher. If the name is NULL, then a name is automatically generated when the launcher starts.

workers

Maximum number of workers to run concurrently when auto-scaling, excluding task retries and manual calls to launch(). Special workers allocated for task retries do not count towards this limit, so the number of workers running at a given time may exceed this maximum. A smaller number of workers may run if the number of executing tasks is smaller than the supplied value of the workers argument.

host

IP address of the mirai client to send and receive tasks. If NULL, the host defaults to nanonext::ip_addr()[1].

port

TCP port to listen for the workers. If NULL, then an available ephemeral port is automatically chosen. Controllers running simultaneously on the same computer (as in a controller group) must not share the same TCP port.

tls

A TLS configuration object from crew_tls().

tls_enable

Deprecated on 2023-09-15 in version 0.4.1. Use argument tls instead.

tls_config

Deprecated on 2023-09-15 in version 0.4.1. Use argument tls instead.

serialization

Either NULL (default) or an object produced by mirai::serial_config() to control the serialization of data sent to workers. This can help with either more efficient data transfers or to preserve attributes of otherwise non-exportable objects (such as torch tensors or arrow tables). See ?mirai::serial_config for details.

profile

Character string, compute profile for mirai::daemons().

seconds_interval

Number of seconds between polling intervals waiting for certain internal synchronous operations to complete. In certain cases, exponential backoff is used with this argument passed to seconds_max in a crew_throttle() object.

seconds_timeout

Number of seconds until timing out while waiting for certain synchronous operations to complete, such as checking mirai::info().

seconds_launch

Seconds of startup time to allow. A worker is unconditionally assumed to be alive from the moment of its launch until seconds_launch seconds later. After seconds_launch seconds, the worker is only considered alive if it is actively connected to its assign websocket.

seconds_idle

Maximum number of seconds that a worker can idle since the completion of the last task. If exceeded, the worker exits. But the timer does not launch until tasks_timers tasks have completed. See the idletime argument of mirai::daemon(). crew does not excel with perfectly transient workers because it does not micromanage the assignment of tasks to workers, so please allow enough idle time for a new worker to be delegated a new task.

seconds_wall

Soft wall time in seconds. The timer does not launch until tasks_timers tasks have completed. See the walltime argument of mirai::daemon().

retry_tasks

Deprecated on 2025-01-13 (crew version 0.10.2.9002).

tasks_max

Maximum number of tasks that a worker will do before exiting. Also determines how often the controller auto-scales. See the Auto-scaling section for details.

tasks_timers

Number of tasks to do before activating the timers for seconds_idle and seconds_wall. See the timerstart argument of mirai::daemon().

reset_globals

TRUE to reset global environment variables between tasks, FALSE to leave them alone.

reset_packages

TRUE to detach any packages loaded during a task (runs between each task), FALSE to leave packages alone. In either case, the namespaces are not detached.

reset_options

TRUE to reset global options to their original state between each task, FALSE otherwise. It is recommended to only set reset_options = TRUE if reset_packages is also TRUE because packages sometimes rely on options they set at loading time. for this and other reasons, reset_options only resets options that were nonempty at the beginning of the task. If your task sets an entirely new option not already in options(), then reset_options = TRUE does not delete the option.

garbage_collection

TRUE to run garbage collection after each task task, FALSE to skip.

crashes_error

Deprecated on 2025-01-13 (crew version 0.10.2.9002).

processes

Deprecated on 2025-08-27 (crew version 1.2.1.9009).

r_arguments

Optional character vector of command line arguments to pass to Rscript (non-Windows) or Rscript.exe (Windows) when starting a worker. Example: r_arguments = c("--vanilla", "--max-connections=32").

crashes_max

In rare cases, a worker may exit unexpectedly before it completes its current task. If this happens, pop() returns a status of "crash" instead of "error" for the task. The controller does not automatically retry the task, but you can retry it manually by calling push() again and using the same task name as before. (However, targets pipelines running crew do automatically retry tasks whose workers crashed.)

crashes_max is a non-negative integer, and it sets the maximum number of allowable consecutive crashes for a given task. If a task's worker crashes more than crashes_max times in a row, then pop() throws an error when it tries to return the results of the task.

backup

An optional crew controller object, or NULL to omit. If supplied, the backup controller runs any pushed tasks that have already reached crashes_max consecutive crashes. Using backup, you can create a chain of controllers with different levels of resources (such as worker memory and CPUs) so that a task that fails on one controller can retry using incrementally more powerful workers. All controllers in a backup chain should be part of the same controller group (see crew_controller_group()) so you can call the group-level pop() and collect() methods to make sure you get results regardless of which controller actually ended up running the task.

Limitations of backup: * crashes_max needs to be positive in order for backup to be used. Otherwise, every task would always skip the current controller and go to backup. * backup cannot be a controller group. It must be an ordinary controller.

options_metrics

Either NULL to opt out of resource metric logging for workers, or an object from crew_options_metrics() to enable and configure resource metric logging for workers. For resource logging to run, the autometric R package version 0.1.0 or higher must be installed.

options_aws_batch

List of options from crew_options_aws_batch(). The job definition and job queue must be specified in crew_options_aws_batch(). crew_options_aws_batch() also allows you to request vCPUs, GPUs, and memory for the jobs.

aws_batch_config