HPC Libs
- José Julián Espina Del Ángel
| Channel | Revision | Published | Runs on |
|---|---|---|---|
| latest/stable | 1 | 03 Jul 2024 |
juju deploy hpc-libs
Deploy universal operators easily with Juju, the Universal Operator Lifecycle Manager.
Platform:
charms.hpc_libs.v0.slurm_ops
-
- Last updated 15 Nov 2024
- Revision Library version 0.8
Abstractions for managing Slurm operations via snap or systemd.
This library contains manager classes that provide high-level interfaces for managing Slurm operations within charmed operators.
Note
This charm library depends on the charms.operator_libs_linux.v0.apt charm library, which can
be imported by running charmcraft fetch-lib charms.operator_libs_linux.v0.apt.
Example Usage
Managing the slurmctld service
The SlurmctldManager class manages the operations of the Slurm controller service.
You can pass the boolean keyword argument snap=True or snap=False to instruct
SlurmctldManager to either use the Slurm snap package or Debian package respectively.
from charms.hpc_libs.v0.slurm_ops import SlurmctldManager
class ApplicationCharm(CharmBase):
# Application charm that needs to use the Slurm snap.
def __init__(self, *args, **kwargs) -> None:
super().__init__(*args, **kwargs)
self._slurm_manager = SlurmctldManager(snap=True)
self.framework.observe(self.on.install, self._on_install)
def _on_install(self, _) -> None:
self._slurmctld.install()
self.unit.set_workload_version(self._slurmctld.version())
with self._slurmctld.config() as config:
config.cluster_name = "cluster"
Index
class SlurmOpsError
Description
Exception raised when a slurm operation failed. None
Methods
SlurmOpsError. message( self )
Description
Return message passed as argument to exception. None
class SlurmctldManager
Description
Manager for the Slurmctld service. None
Methods
SlurmctldManager. __init__( self )
class SlurmdManager
Manager for the Slurmd service.
Description
This service will additionally provide some environment variables that need to be passed through to the service in case the default service is overriden (e.g. a systemctl file override).
- SLURMD_CONFIG_SERVER. Sets the `--conf-server` option for `slurmd`.Methods
SlurmdManager. __init__( self )
SlurmdManager. user( self )
Description
Get the SlurmdUser. None
SlurmdManager. group( self )
Description
Get the SlurmdUser group. None
SlurmdManager. config_server( self )
Description
Get the config server address of this Slurmd node. None
SlurmdManager. config_server( self , addr: str )
Description
Set the config server address of this Slurmd node. None
SlurmdManager. config_server( self )
Description
Unset the config server address of this Slurmd node. None
class SlurmdbdManager
Description
Manager for the Slurmdbd service. None
Methods
SlurmdbdManager. __init__( self )
SlurmdbdManager. mysql_unix_port( self )
Description
Get the URI of the unix socket slurmdbd uses to communicate with MySQL. None
SlurmdbdManager. mysql_unix_port( self , socket_path )
Description
Set the unix socket URI that slurmdbd will use to communicate with MySQL. None
SlurmdbdManager. mysql_unix_port( self )
Description
Delete the configured unix socket URI. None
class SlurmrestdManager
Description
Manager for the Slurmrestd service. None
Methods
SlurmrestdManager. __init__( self )
SlurmrestdManager. user( self )
Description
Get the user that the slurmrestd service will run as. None
SlurmrestdManager. group( self )
Description
Get the group that the slurmrestd service will run as. None