Skip to main content

H2O Engine profiles

H2O Engine profiles extend the common profile fields with node count constraints for distributed processing across multiple H2O-3 nodes. For background on how profiles work, see Engine profiles.

Profile fields​

The following table lists the fields specific to H2O Engine profiles. For shared fields (CPU, GPU, memory, timeouts, and OIDC roles), see the shared profile fields table.

FieldRequiredTypeDescription
node_count_constraintRequiredNumeric constraintAllowed range for the number of H2O-3 nodes in a distributed cluster

Node count constraint​

H2O-3 supports distributed processing across multiple nodes. The node_count_constraint field defines how many nodes a user can allocate. CPU, GPU, and memory are allocated per node, so total resources equal per-node values multiplied by the node count.

tip

When sizing the node count constraint, consider cluster capacity. A profile that allows up to 5 nodes with 16 CPUs each consumes 80 CPUs per engine. Set a cumulative_maximum on the node count constraint to cap the total number of nodes a single user can run across all their engines.

Manage H2O Engine profiles with the Python client​

For installation instructions, see Python client installation.

List available profiles​

To list profiles assigned to you based on your OIDC roles:

import h2o_engine_manager

aiem = h2o_engine_manager.login()
h2o_engine_profile_client = aiem.h2o_engine_profile_client

profiles = h2o_engine_profile_client.list_all_assigned_h2o_engine_profiles(
    parent="workspaces/global"
)
for profile in profiles:
    print(f"{profile.name}: {profile.display_name}")

Create a profile (administrator)​

from h2o_engine_manager.clients.h2o_engine_profile.h2o_engine_profile import H2OEngineProfile
from h2o_engine_manager.clients.constraint.profile_constraint_numeric import ProfileConstraintNumeric
from h2o_engine_manager.clients.constraint.profile_constraint_duration import ProfileConstraintDuration

profile = H2OEngineProfile(
    display_name="ML Engineering - Multi-node",
    priority=1,
    enabled=True,
    assigned_oidc_roles_enabled=True,
    assigned_oidc_roles=["ml-engineering"],
    node_count_constraint=ProfileConstraintNumeric(
        minimum="1", default="1", maximum="5", cumulative_maximum="10"
    ),
    cpu_constraint=ProfileConstraintNumeric(minimum="1", default="4", maximum="16"),
    gpu_constraint=ProfileConstraintNumeric(minimum="0", default="0", maximum="2"),
    memory_bytes_constraint=ProfileConstraintNumeric(
        minimum="4Gi", default="16Gi", maximum="64Gi"
    ),
    max_idle_duration_constraint=ProfileConstraintDuration(
        minimum="30m", default="1h", maximum="8h"
    ),
    max_running_duration_constraint=ProfileConstraintDuration(
        minimum="1h", default="8h", maximum="1d"
    ),
)

created = h2o_engine_profile_client.create_h2o_engine_profile(
    parent="workspaces/global",
    h2o_engine_profile=profile,
    h2o_engine_profile_id="ml-eng-multinode",
)
print(f"Created profile: {created.name}")

Update a profile (administrator)​

To update specific fields of an existing profile, retrieve it first, modify the fields, and pass an update_mask specifying which fields to update:

profile = h2o_engine_profile_client.get_h2o_engine_profile(
    name="workspaces/global/h2oEngineProfiles/ml-eng-multinode"
)
profile.node_count_constraint = ProfileConstraintNumeric(
    minimum="1", default="2", maximum="8", cumulative_maximum="16"
)

updated = h2o_engine_profile_client.update_h2o_engine_profile(
    h2o_engine_profile=profile,
    update_mask="node_count_constraint",
)
print(f"Updated profile: {updated.name}")

Pass update_mask="*" to update all fields at once.

Delete a profile (administrator)​

h2o_engine_profile_client.delete_h2o_engine_profile(
    name="workspaces/global/h2oEngineProfiles/ml-eng-multinode"
)

Create an engine with a profile​

CPU, GPU, and memory values are per node. With node_count=3 and cpu=8, the engine uses 24 total CPU units.

profiles = h2o_engine_profile_client.list_all_assigned_h2o_engine_profiles(
    parent="workspaces/global"
)
selected_profile = profiles[0]

engine = aiem.h2o_engine_client.create_engine(
    workspace_id="default",
    engine_id="my-h2o-engine",
    display_name="My H2O Engine",
    profile=selected_profile.name,
    h2o_engine_version="workspaces/global/h2oEngineVersions/3.46.0.6",
    cpu=8,              # per node
    gpu=0,              # per node
    memory_bytes="16Gi",  # per node
    node_count=1,
    max_idle_duration="1h",
    max_running_duration="8h",
)
Feedback