OARCluster(n_workers=0, job_cls: dask_jobqueue.core.Job = None, loop=None, security=None, silence_logs='error', name=None, asynchronous=False, interface=None, host=None, protocol='tcp://', dashboard_address=':8787', config_name=None, **kwargs)¶
Launch Dask on an OAR cluster
Destination queue for each worker job. Passed to #OAR -q option.
Accounting string associated with each worker job. Passed to #OAR -p option.
Total number of cores per job
- memory: str
Total amount of memory per job
Cut the job up into this many processes. Good for GIL workloads or for nodes with many cores. By default,
process ~= sqrt(cores)so that the number of processes and the number of threads per process is roughly the same.
Network interface like ‘eth0’ or ‘ib0’.
Whether or not to start a nanny process
Dask worker local directory for file spilling.
Seconds to wait for a scheduler before closing workers
Additional arguments to pass to dask-worker
Other commands to add to script before launching worker.
Lines to skip in the header. Header lines matching this text will be removed
Directory to use for job scheduler logs.
Path to desired interpreter for your batch submission script.
Python executable used to launch Dask workers. Defaults to the Python that is submitting these jobs
Section to use from jobqueue.yaml configuration file.
Name of Dask worker. This is typically set by the Cluster
Number of workers to start by default. Defaults to 0. See the scale method
Log level like “debug”, “info”, or “error” to emit here if the scheduler is started locally
Whether or not to run this cluster object with the async/await syntax
A dask.distributed security object if you’re using TLS/SSL
- dashboard_addressstr or int
An address like “:8787” on which to host the Scheduler’s dashboard
Request resources and specify job placement. Passed to #OAR -l option.
Walltime for each worker job.
List of other OAR options, for example -t besteffort. Each option will be prepended with the #OAR prefix.
>>> from dask_jobqueue import OARCluster >>> cluster = OARCluster(queue='regular') >>> cluster.scale(jobs=10) # ask for 10 jobs
>>> from dask.distributed import Client >>> client = Client(cluster)
This also works with adaptive clusters. This automatically launches and kill workers based on load.
__init__(self, n_workers=0, job_cls:dask_jobqueue.core.Job=None, loop=None, security=None, silence_logs='error', name=None, asynchronous=False, interface=None, host=None, protocol='tcp://', dashboard_address=':8787', config_name=None, **kwargs)¶
Initialize self. See help(type(self)) for accurate signature.
__init__(self[, n_workers, loop, security, …])
adapt(self, \*args, minimum_jobs, …)
Scale Dask cluster automatically based on scheduler activity.
logs(self[, scheduler, workers])
Return logs for the scheduler and workers
Return name and spec for the next worker
scale(self[, n, jobs, memory, cores])
Scale cluster to specified configurations.
scale_up(self[, n, memory, cores])
Scale cluster to n workers
sync(self, func, \*args[, asynchronous, …])