buildgrid.server.scheduler.impl module

class buildgrid.server.scheduler.impl.SchedulerMetrics

Bases: TypedDict

jobs: dict[tuple[str, str], int]

class buildgrid.server.scheduler.impl.AgedJobHandlerOptions(job_max_age, handling_period, max_handling_window)

Bases: NamedTuple

job_max_age: timedelta: Alias for field number 0

handling_period: timedelta: Alias for field number 1

max_handling_window: int: Alias for field number 2

static from_config(job_max_age_cfg: dict[str, float], handling_period_cfg: dict[str, float] | None = None, max_handling_window_cfg: int | None = None) → AgedJobHandlerOptions: Helper method for creating AgedJobHandlerOptions objects If input configs are None, assign defaults

class buildgrid.server.scheduler.impl.Scheduler(sql_provider: SqlProvider, storage: StorageABC, *, sql_ro_provider: SqlProvider | None = None, sql_notifier_provider: SqlProvider | None = None, property_set: PropertySet, action_cache: ActionCacheABC | None = None, action_browser_url: str | None = None, max_execution_timeout: int = 7200, metering_client: SyncMeteringServiceClient | None = None, metering_throttle_action: MeteringThrottleAction | None = None, bot_session_keepalive_timeout: int = 600, logstream_channel: Channel | None = None, logstream_instance: str | None = None, asset_client: AssetClient | None = None, queued_action_retention_hours: float | None = None, completed_action_retention_hours: float | None = None, action_result_retention_hours: float | None = None, enable_job_watcher: bool = False, poll_interval: float = 1, pruning_options: AgedJobHandlerOptions | None = None, queue_timeout_options: AgedJobHandlerOptions | None = None, max_job_attempts: int = 5, job_assignment_interval: float = 1.0, priority_assignment_percentage: int = 100, max_queue_size: int | None = None, execution_timer_interval: float = 60.0, session_expiry_timer_interval: float = 10.0)

Bases: object

RETRYABLE_STATUS_CODES = (13, 14)

start() → None

stop() → None

queue_job_action(*, action: Action, action_digest: Digest, command: Command, platform_requirements: dict[str, list[str]], property_label: str, priority: int, skip_cache_lookup: bool, request_metadata: RequestMetadata | None = None, client_identity: ClientIdentityEntry | None = None) → str: De-duplicates or inserts a newly created job into the execution queue. Returns an operation name associated with this job.

create_operation_for_existing_job(*, action_digest: Digest, priority: int, request_metadata: RequestMetadata | None, client_identity: ClientIdentityEntry | None) → str | None

create_operation_for_new_job(*, action: Action, action_digest: Digest, command: Command, execute_response: ExecuteResponse | None, platform_requirements: dict[str, list[str]], property_label: str, priority: int, request_metadata: RequestMetadata | None = None, client_identity: ClientIdentityEntry | None = None) → str

create_operation(job_name: str, *, request_metadata: RequestMetadata | None = None, client_identity: ClientIdentityEntry | None = None) → str

load_operation(operation_name: str) → Operation

get_operation_job_name(operation_name: str) → str | None

get_operation_request_metadata_by_name(operation_name: str) → RequestMetadata | None

get_client_identity_by_operation(operation_name: str) → ClientIdentity | None

execution_timer_loop(shutdown_requested: Event) → None: Periodically timeout aged executing jobs

cancel_jobs_exceeding_execution_timeout(max_execution_timeout: int | None = None) → None

cancel_operation(operation_name: str) → None

list_operations(operation_filters: list[buildgrid.server.operations.filtering.filter.OperationFilter] | None = None, page_size: int | None = None, page_token: str | None = None) → tuple[list[buildgrid._protos.google.longrunning.operations_pb2.Operation], str]

list_workers(name_filter: str, page_number: int, page_size: int) → tuple[list[buildgrid.server.sql.models.BotEntry], int]

get_metrics() → SchedulerMetrics | None

assign_n_leases_by_priority(*, capability_hash: str, bot_names: list[str]) → list[str]

assign_n_leases_by_age(*, capability_hash: str, bot_names: list[str]) → list[str]

queue_timer_loop(shutdown_requested: Event) → None: Periodically timeout aged queued jobs

prune_timer_loop(shutdown_requested: Event) → None: Running in a background thread, this method wakes up periodically and deletes older records from the jobs tables using configurable parameters

get_or_create_client_identity_in_store(session: Session, client_id: ClientIdentityEntry) → ClientIdentityEntry

Get the ClientIdentity in the storage or create one. This helper function essentially makes sure the client_id is created during the transaction

Parameters:

session (Session) – sqlalchemy Session
client_id (ClientIdentityEntry) – identity of the client that creates an operation

Returns:

identity of the client that creates an operation

Return type:

ClientIdentityEntry

get_or_create_request_metadata_in_store(session: Session, request_metadata: RequestMetadata) → RequestMetadataEntry

add_bot_entry(*, bot_session_id: str, bot_session_status: int) → str

close_bot_sessions(bot_name: str) → None

session_expiry_timer_loop(shutdown_requested: Event) → None

reap_expired_sessions() → bool: Find and close expired bot sessions. Returns True if sessions were closed. Only closes a few sessions to minimize time in transaction.

synchronize_bot_lease(bot_name: str, bot_id: str, bot_status: int, session_lease: Lease | None, partial_execution_metadata: dict[str, buildgrid._protos.build.bazel.remote.execution.v2.remote_execution_pb2.ExecutedActionMetadata] | None = None) → Lease | None

count_bots_by_status() → dict[buildgrid.server.enums.BotStatus, int]: Count the number of bots with a particular status

refresh_bot_expiry_time(bot_name: str, bot_id: str) → datetime

This update is done out-of-band from the main synchronize_bot_lease transaction, as there are cases where we will skip calling the synchronization, but still want the session to be updated such that it does not get reaped. This slightly duplicates the update happening in synchronize_bot_lease, however, that update is still required to not have the job reaped during its job assignment waiting period.

This method should be called at the end of the update and create bot session methods. The returned datetime should be assigned to the deadline within the returned session proto.

get_metadata_for_leases(leases: Iterable[Lease]) → list[tuple[str, bytes]]

Return a list of Job metadata for a given list of leases.

Parameters:: leases (list) – List of leases to get Job metadata for.
Returns:: List of tuples of the form ('executeoperationmetadata-bin': serialized_metadata).

get_execute_action_metadata(job: JobEntry) → ExecutedActionMetadata

publish_execution_stats(job_name: str, execution_metadata: ExecutedActionMetadata, property_label: str = 'unknown') → None