hipscatalog_gen.cluster package
Submodules
hipscatalog_gen.cluster.runtime module
Dask cluster setup and teardown utilities.
- class ClusterRuntime(cluster, client, diagnostics_mode)[source]
Bases:
objectRuntime handles for the Dask cluster.
- Parameters:
cluster (Any)
client (Client)
diagnostics_mode (str)
- cluster: Any
- client: Client
- diagnostics_mode: str
- setup_cluster(cfg, report_dir, log_fn)[source]
Create and configure the Dask cluster and diagnostics context.
- Parameters:
cfg (ClusterCfg) – Cluster configuration (local or SLURM) with worker counts and memory limits.
report_dir (Path) – Directory where per-step diagnostics reports are written.
log_fn (Callable[[str, bool], None]) – Logging callback
(message, always).
- Returns:
A pair
(runtime, diag_ctx_factory), where:runtime:ClusterRuntimewith cluster/client handles and flags.diag_ctx_factory: callablelabel -> context managerused aswith diag_ctx_factory("step_name"):around pipeline steps.
- Return type:
Tuple[ClusterRuntime, Callable[[str], ContextManager[Any]]]
- Raises:
ImportError – If
mode='slurm'is set butdask-jobqueueis not available.
- shutdown_cluster(runtime)[source]
Gracefully close client and cluster.
- Parameters:
runtime (ClusterRuntime) – ClusterRuntime with cluster and client handles.
- Return type:
None
Module contents
Cluster orchestration helpers for hipscatalog-gen.
- class ClusterRuntime(cluster, client, diagnostics_mode)[source]
Bases:
objectRuntime handles for the Dask cluster.
- Parameters:
cluster (Any)
client (Client)
diagnostics_mode (str)
- cluster: Any
- client: Client
- diagnostics_mode: str
- setup_cluster(cfg, report_dir, log_fn)[source]
Create and configure the Dask cluster and diagnostics context.
- Parameters:
cfg (ClusterCfg) – Cluster configuration (local or SLURM) with worker counts and memory limits.
report_dir (Path) – Directory where per-step diagnostics reports are written.
log_fn (Callable[[str, bool], None]) – Logging callback
(message, always).
- Returns:
A pair
(runtime, diag_ctx_factory), where:runtime:ClusterRuntimewith cluster/client handles and flags.diag_ctx_factory: callablelabel -> context managerused aswith diag_ctx_factory("step_name"):around pipeline steps.
- Return type:
Tuple[ClusterRuntime, Callable[[str], ContextManager[Any]]]
- Raises:
ImportError – If
mode='slurm'is set butdask-jobqueueis not available.
- shutdown_cluster(runtime)[source]
Gracefully close client and cluster.
- Parameters:
runtime (ClusterRuntime) – ClusterRuntime with cluster and client handles.
- Return type:
None