Migrate to crew 1.0.0

crew.cluster 0.3.4

• Deprecate retryable options because crew 1.0.0 moved away from them.

Retryable options and integration with autometric

crew.cluster 0.3.3

• Add slurm_memory_gigabytes_required to set --mem in SLURM (#44, @multimeric).
• Add r_arguments to accept R command line arguments for workers.
• Support options_metrics.
• Organize arguments into options lists.
• Use crashes_error from crew.
• Implement retryable options for memory, cores, GPUs, wall time, and SLURM partition (#48, @stemangiola).
• Set default seconds_idle to 300.

retry_tasks

crew.cluster 0.3.2

• Add retry_tasks argument.

SLURM monitor

crew.cluster 0.3.1

• Add a monitor class for SLURM clusters.

Abstract monitor class and higher seconds_timeout

crew.cluster 0.3.0

• Create an abstract monitor class for cluster-specific monitor classes to inherit from (#32).
• Require crew >= 0.8.0.
• Raise seconds_timeout to 60 across controllers.

SGE monitor class

crew.cluster 0.2.0

• Re-enable unit tests.
• Require crew >= 0.7.0.
• Add a "monitor" class for SGE clusters to conveniently list and terminate jobs.
• Deprecate command_delete in favor of command_terminate.

Encapsulation

crew.cluster 0.1.4

• Refactor methods for development crew.
• Encapsulate non-function R6 members in the private list.

Minor tweaks

crew.cluster 0.1.3

• Deprecate seconds_exit (wlandau/crew#125, @shikokuchuo).
• Deprecate seconds_interval (wlandau/crew#131).
• Add a new slurm_partition argument the SLURM controller and launcher (#24, @kkmann).
• Turn on automatic TLS encryption by default.

TLS and SLURM configuration

crew.cluster 0.1.2

• Require crew >= 0.5.0.
• Add a new slurm_time_minutes for SLURM wall time (#1, @cfljam).
• Enable TLS.
• Use MB format for memory in SLURM launcher (#22, @kendonB).
• Use launch_max and crew_tls() from crew 0.5.0.

Compatibility with upcoming changes in crew

crew.cluster 0.1.1

• Align with breaking changes in crew 0.3.0.
• Increase seconds_launch to 86400 (1 day).