Warning
This document is for an old release of Galaxy. You can alternatively view this page in the latest release if it exists or view the top of the latest release's documentation.
Source code for galaxy.jobs.runners.pulsar
"""Job runner used to execute Galaxy jobs through Pulsar.
More information on Pulsar can be found at https://pulsar.readthedocs.io/ .
"""
import copy
import errno
import logging
import os
import re
import subprocess
from time import sleep
from typing import (
Any,
Dict,
Optional,
)
import pulsar.core
import yaml
from packaging.version import Version
from pulsar.client import (
build_client_manager,
CLIENT_INPUT_PATH_TYPES,
ClientInput,
ClientInputs,
ClientJobDescription,
ClientOutputs,
EXTENDED_METADATA_DYNAMIC_COLLECTION_PATTERN,
finish_job as pulsar_finish_job,
PathMapper,
PulsarClientTransportError,
PulsarOutputs,
submit_job as pulsar_submit_job,
url_to_destination_params,
)
# TODO: Perform pulsar release with this included in the client package
from pulsar.client.staging import DEFAULT_DYNAMIC_COLLECTION_PATTERN
from galaxy import model
from galaxy.job_execution.compute_environment import (
ComputeEnvironment,
dataset_path_to_extra_path,
)
from galaxy.jobs import JobDestination
from galaxy.jobs.command_factory import build_command
from galaxy.jobs.runners import (
AsynchronousJobRunner,
AsynchronousJobState,
JobState,
)
from galaxy.tool_util.deps import dependencies
from galaxy.util import (
galaxy_directory,
specs,
string_as_bool_or_none,
)
log = logging.getLogger(__name__)
__all__ = (
"PulsarLegacyJobRunner",
"PulsarRESTJobRunner",
"PulsarMQJobRunner",
"PulsarEmbeddedJobRunner",
"PulsarEmbeddedMQJobRunner",
)
MINIMUM_PULSAR_VERSIONS = {
"_default_": Version("0.7.0.dev3"),
"remote_metadata": Version("0.8.0"),
"remote_container_handling": Version("0.9.1.dev0"), # probably 0.10 ultimately?
}
NO_REMOTE_GALAXY_FOR_METADATA_MESSAGE = "Pulsar misconfiguration - Pulsar client configured to set metadata remotely, but remote Pulsar isn't properly configured with a galaxy_home directory."
NO_REMOTE_DATATYPES_CONFIG = "Pulsar client is configured to use remote datatypes configuration when setting metadata externally, but Pulsar is not configured with this information. Defaulting to datatypes_conf.xml."
GENERIC_REMOTE_ERROR = "Failed to communicate with remote job server."
FAILED_REMOTE_ERROR = "Remote job server indicated a problem running or monitoring this job."
LOST_REMOTE_ERROR = "Remote job server could not determine this job's state."
UPGRADE_PULSAR_ERROR = "Galaxy is misconfigured, please contact administrator. The target Pulsar server is unsupported, this version of Galaxy requires Pulsar version %s or newer."
# Is there a good way to infer some default for this? Can only use
# url_for from web threads. https://gist.github.com/jmchilton/9098762
DEFAULT_GALAXY_URL = "http://localhost:8080"
PULSAR_PARAM_SPECS = dict(
transport=dict(map=specs.to_str_or_none, valid=specs.is_in("urllib", "curl", None), default=None),
transport_timeout=dict(
map=lambda val: None if val == "None" else int(val),
default=None,
),
cache=dict(
map=specs.to_bool_or_none,
default=None,
),
remote_container_handling=dict(
map=specs.to_bool,
default=False,
),
amqp_url=dict(
map=specs.to_str_or_none,
default=None,
),
amqp_key_prefix=dict(
map=specs.to_str_or_none,
default=None,
),
galaxy_url=dict(
map=specs.to_str_or_none,
default=None,
),
secret=dict(
map=specs.to_str_or_none,
default=None,
),
pulsar_config=dict(
map=specs.to_str_or_none,
default=None,
),
pulsar_app_config=dict(
default=None,
),
manager=dict(
map=specs.to_str_or_none,
default=None,
),
persistence_directory=dict(
map=specs.to_str_or_none,
default=None,
),
amqp_acknowledge=dict(map=specs.to_bool_or_none, default=None),
amqp_ack_republish_time=dict(
map=lambda val: None if val == "None" else int(val),
default=None,
),
amqp_consumer_timeout=dict(
map=lambda val: None if val == "None" else float(val),
default=None,
),
amqp_connect_ssl_ca_certs=dict(
map=specs.to_str_or_none,
default=None,
),
amqp_connect_ssl_keyfile=dict(
map=specs.to_str_or_none,
default=None,
),
amqp_connect_ssl_certfile=dict(
map=specs.to_str_or_none,
default=None,
),
amqp_connect_ssl_cert_reqs=dict(
map=specs.to_str_or_none,
default=None,
),
# https://kombu.readthedocs.io/en/latest/reference/kombu.html#kombu.Producer.publish
amqp_publish_retry=dict(
map=specs.to_bool,
default=False,
),
amqp_publish_priority=dict(
map=int,
valid=lambda x: 0 <= x and x <= 9,
default=0,
),
# https://kombu.readthedocs.io/en/latest/reference/kombu.html#kombu.Exchange.delivery_mode
amqp_publish_delivery_mode=dict(
map=str,
valid=specs.is_in("transient", "persistent"),
default="persistent",
),
amqp_publish_retry_max_retries=dict(
map=int,
default=None,
),
amqp_publish_retry_interval_start=dict(
map=int,
default=None,
),
amqp_publish_retry_interval_step=dict(
map=int,
default=None,
),
amqp_publish_retry_interval_max=dict(
map=int,
default=None,
),
)
PARAMETER_SPECIFICATION_REQUIRED = object()
PARAMETER_SPECIFICATION_IGNORED = object()
class PulsarJobRunner(AsynchronousJobRunner):
"""Base class for pulsar job runners."""
start_methods = ["_init_worker_threads", "_init_client_manager", "_monitor"]
runner_name = "PulsarJobRunner"
default_build_pulsar_app = False
use_mq = False
poll = True
def __init__(self, app, nworkers, **kwds):
"""Start the job runner."""
super().__init__(app, nworkers, runner_param_specs=PULSAR_PARAM_SPECS, **kwds)
galaxy_url = self.runner_params.galaxy_url
if not galaxy_url:
galaxy_url = app.config.galaxy_infrastructure_url
if galaxy_url:
galaxy_url = galaxy_url.rstrip("/")
self.galaxy_url = galaxy_url
def _monitor(self):
if self.use_mq:
# This is a message queue driven runner, don't monitor
# just setup required callback.
self.client_manager.ensure_has_status_update_callback(self.__async_update)
self.client_manager.ensure_has_ack_consumers()
if self.poll:
self._init_monitor_thread()
else:
self._init_noop_monitor()
def _init_client_manager(self):
pulsar_conf = self.runner_params.get("pulsar_app_config", None)
pulsar_conf_file = None
if pulsar_conf is None:
pulsar_conf_file = self.runner_params.get("pulsar_config", None)
self.__init_pulsar_app(pulsar_conf, pulsar_conf_file)
client_manager_kwargs = {}
for kwd in "manager", "cache", "transport", "persistence_directory":
client_manager_kwargs[kwd] = self.runner_params[kwd]
if self.pulsar_app is not None:
client_manager_kwargs["pulsar_app"] = self.pulsar_app
# TODO: Hack remove this following line pulsar lib update
# that includes https://github.com/galaxyproject/pulsar/commit/ce0636a5b64fae52d165bcad77b2caa3f0e9c232
client_manager_kwargs["file_cache"] = None
for kwd in self.runner_params.keys():
if kwd.startswith("amqp_") or kwd.startswith("transport_"):
client_manager_kwargs[kwd] = self.runner_params[kwd]
self.client_manager = build_client_manager(**client_manager_kwargs)
def __init_pulsar_app(self, conf, pulsar_conf_path):
if conf is None and pulsar_conf_path is None and not self.default_build_pulsar_app:
self.pulsar_app = None
return
if conf is None:
conf = {}
if pulsar_conf_path is None:
log.info("Creating a Pulsar app with default configuration (no pulsar_conf specified).")
else:
log.info(f"Loading Pulsar app configuration from {pulsar_conf_path}")
with open(pulsar_conf_path) as f:
conf.update(yaml.safe_load(f) or {})
if "job_metrics_config_file" not in conf:
conf["job_metrics"] = self.app.job_metrics
if "staging_directory" not in conf:
conf["staging_directory"] = os.path.join(self.app.config.data_dir, "pulsar_staging")
if "persistence_directory" not in conf:
conf["persistence_directory"] = os.path.join(self.app.config.data_dir, "pulsar_persisted_data")
if "galaxy_home" not in conf:
conf["galaxy_home"] = galaxy_directory()
self.pulsar_app = pulsar.core.PulsarApp(**conf)
def url_to_destination(self, url):
"""Convert a legacy URL to a job destination."""
return JobDestination(runner="pulsar", params=url_to_destination_params(url))
def check_watched_item(self, job_state):
if self.use_mq:
# Might still need to check pod IPs.
job_wrapper = job_state.job_wrapper
guest_ports = job_wrapper.guest_ports
if len(guest_ports) > 0:
persisted_state = job_wrapper.get_state()
if persisted_state in model.Job.terminal_states + [model.Job.states.DELETING]:
log.debug(
"(%s) Watched job in terminal state, will stop monitoring: %s",
job_state.job_id,
persisted_state,
)
job_state = None
elif persisted_state == model.Job.states.RUNNING:
client = self.get_client_from_state(job_state)
job_ip = client.job_ip()
if job_ip:
ports_dict = {}
for guest_port in guest_ports:
ports_dict[str(guest_port)] = dict(host=job_ip, port=guest_port, protocol="http")
self.app.interactivetool_manager.configure_entry_points(job_wrapper.get_job(), ports_dict)
log.debug("(%s) Got ports for entry point: %s", job_state.job_id, str(ports_dict))
job_state = None
else:
# No need to monitor MQ jobs that have no entry points
job_state = None
return job_state
else:
return self.check_watched_item_state(job_state)
def check_watched_item_state(self, job_state):
try:
client = self.get_client_from_state(job_state)
status = client.get_status()
except PulsarClientTransportError as exc:
log.error("Communication error with Pulsar server on state check, will retry: %s", exc)
return job_state
except Exception:
# An orphaned job was put into the queue at app startup, so remote server went down
# either way we are done I guess.
self.mark_as_finished(job_state)
return None
job_state = self._update_job_state_for_status(job_state, status)
return job_state
def _update_job_state_for_status(self, job_state, pulsar_status, full_status=None):
log.debug("(%s) Received status update: %s", job_state.job_id, pulsar_status)
if pulsar_status in ["complete", "cancelled"] or job_state.job_wrapper.get_state() == model.Job.states.STOPPED:
self.mark_as_finished(job_state)
return None
if pulsar_status in ["failed", "lost"]:
if pulsar_status == "failed":
message = FAILED_REMOTE_ERROR
else:
message = LOST_REMOTE_ERROR
if not job_state.job_wrapper.get_job().finished:
self.fail_job(job_state, message=message, full_status=full_status)
return None
if pulsar_status == "running" and not job_state.running:
job_state.running = True
job_state.job_wrapper.change_state(model.Job.states.RUNNING)
return job_state
def queue_job(self, job_wrapper):
job_destination = job_wrapper.job_destination
self._populate_parameter_defaults(job_destination)
command_line, client, remote_job_config, compute_environment, remote_container = self.__prepare_job(
job_wrapper, job_destination
)
if not command_line:
return
try:
dependencies_description = PulsarJobRunner.__dependencies_description(client, job_wrapper)
rewrite_paths = not PulsarJobRunner.__rewrite_parameters(client)
path_rewrites_unstructured = {}
output_names = []
if compute_environment:
path_rewrites_unstructured = compute_environment.path_rewrites_unstructured
output_names = compute_environment.output_names()
client_inputs_list = []
for input_dataset_wrapper in job_wrapper.job_io.get_input_paths():
# str here to resolve false_path if set on a DatasetPath object.
path = str(input_dataset_wrapper)
object_store_ref = {
"dataset_id": input_dataset_wrapper.dataset_id,
"dataset_uuid": str(input_dataset_wrapper.dataset_uuid),
"object_store_id": input_dataset_wrapper.object_store_id,
}
client_inputs_list.append(
ClientInput(path, CLIENT_INPUT_PATH_TYPES.INPUT_PATH, object_store_ref=object_store_ref)
)
for input_extra_path in compute_environment.path_rewrites_input_extra.keys():
# TODO: track dataset for object_Store_ref...
client_inputs_list.append(
ClientInput(input_extra_path, CLIENT_INPUT_PATH_TYPES.INPUT_EXTRA_FILES_PATH)
)
for input_metadata_path in compute_environment.path_rewrites_input_metadata.keys():
# TODO: track dataset for object_Store_ref...
client_inputs_list.append(
ClientInput(input_metadata_path, CLIENT_INPUT_PATH_TYPES.INPUT_METADATA_PATH)
)
input_files = None
client_inputs = ClientInputs(client_inputs_list)
else:
input_files = self.get_input_files(job_wrapper)
client_inputs = None
if self.app.config.metadata_strategy == "legacy":
# Drop this branch in 19.09.
metadata_directory = job_wrapper.working_directory
else:
metadata_directory = os.path.join(job_wrapper.working_directory, "metadata")
dest_params = job_destination.params
remote_pulsar_app_config = dest_params.get("pulsar_app_config", {}).copy()
if "pulsar_app_config_path" in dest_params:
pulsar_app_config_path = dest_params["pulsar_app_config_path"]
with open(pulsar_app_config_path) as fh:
remote_pulsar_app_config.update(yaml.safe_load(fh))
job_directory_files = []
config_files = job_wrapper.extra_filenames
tool_script = os.path.join(job_wrapper.working_directory, "tool_script.sh")
if os.path.exists(tool_script):
log.debug(f"Registering tool_script for Pulsar transfer [{tool_script}]")
job_directory_files.append(tool_script)
config_files.append(tool_script)
# Following is job destination environment variables
env = client.env
# extend it with tool defined environment variables
tool_envs = job_wrapper.environment_variables
env.extend(tool_envs)
for tool_env in tool_envs:
job_directory_path = tool_env.get("job_directory_path")
if job_directory_path:
config_files.append(job_directory_path)
tool_directory_required_files = job_wrapper.tool.required_files
client_job_description = ClientJobDescription(
command_line=command_line,
input_files=input_files,
client_inputs=client_inputs, # Only one of these input defs should be non-None
client_outputs=self.__client_outputs(client, job_wrapper),
working_directory=job_wrapper.tool_working_directory,
metadata_directory=metadata_directory,
tool=job_wrapper.tool,
config_files=config_files,
dependencies_description=dependencies_description,
env=env,
rewrite_paths=rewrite_paths,
arbitrary_files=path_rewrites_unstructured,
touch_outputs=output_names,
remote_pulsar_app_config=remote_pulsar_app_config,
job_directory_files=job_directory_files,
container=None if not remote_container else remote_container.container_id,
guest_ports=job_wrapper.guest_ports,
tool_directory_required_files=tool_directory_required_files,
)
external_job_id = pulsar_submit_job(client, client_job_description, remote_job_config)
log.info(f"Pulsar job submitted with job_id {external_job_id}")
job = job_wrapper.get_job()
# Set the job destination here (unlike other runners) because there are likely additional job destination
# params from the Pulsar client.
# Flush with change_state.
job_wrapper.set_job_destination(job_destination, external_id=external_job_id, flush=False, job=job)
job_wrapper.change_state(model.Job.states.QUEUED, job=job)
except Exception:
job_wrapper.fail("failure running job", exception=True)
log.exception("failure running job %d", job_wrapper.job_id)
return
pulsar_job_state = AsynchronousJobState(
job_wrapper=job_wrapper, job_id=external_job_id, job_destination=job_destination
)
pulsar_job_state.old_state = True
pulsar_job_state.running = False
self.monitor_job(pulsar_job_state)
def __needed_features(self, client):
return {
"remote_metadata": PulsarJobRunner.__remote_metadata(client),
"remote_container_handling": PulsarJobRunner.__remote_container_handling(client),
}
def __prepare_job(self, job_wrapper, job_destination):
"""Build command-line and Pulsar client for this job."""
command_line = None
client = None
remote_job_config = None
compute_environment: Optional[PulsarComputeEnvironment] = None
remote_container = None
fail_or_resubmit = False
try:
client = self.get_client_from_wrapper(job_wrapper)
tool = job_wrapper.tool
remote_job_config = client.setup(tool.id, tool.version, tool.requires_galaxy_python_environment)
remote_container_handling = PulsarJobRunner.__remote_container_handling(client)
if remote_container_handling:
# Handle this remotely and don't pass it to build_command
remote_container = self._find_container(
job_wrapper,
)
needed_features = self.__needed_features(client)
PulsarJobRunner.check_job_config(remote_job_config, check_features=needed_features)
rewrite_parameters = PulsarJobRunner.__rewrite_parameters(client)
prepare_kwds = {}
if rewrite_parameters:
compute_environment = PulsarComputeEnvironment(client, job_wrapper, remote_job_config)
prepare_kwds["compute_environment"] = compute_environment
job_wrapper.prepare(**prepare_kwds)
self.__prepare_input_files_locally(job_wrapper)
remote_metadata = PulsarJobRunner.__remote_metadata(client)
dependency_resolution = PulsarJobRunner.__dependency_resolution(client)
metadata_kwds = self.__build_metadata_configuration(
client,
job_wrapper,
remote_metadata,
remote_job_config,
compute_environment=compute_environment,
)
remote_working_directory = remote_job_config["working_directory"]
remote_job_directory = os.path.abspath(os.path.join(remote_working_directory, os.path.pardir))
remote_tool_directory = os.path.abspath(os.path.join(remote_job_directory, "tool_files"))
pulsar_version = PulsarJobRunner.pulsar_version(remote_job_config)
remote_command_params = dict(
working_directory=remote_job_config["metadata_directory"],
script_directory=remote_job_directory,
metadata_kwds=metadata_kwds,
dependency_resolution=dependency_resolution,
pulsar_version=pulsar_version,
)
rewrite_paths = not PulsarJobRunner.__rewrite_parameters(client)
if pulsar_version < Version("0.14.999") and rewrite_paths:
job_wrapper.disable_commands_in_new_shell()
container = None
if remote_container is None:
container = self._find_container(
job_wrapper,
compute_working_directory=remote_working_directory,
compute_tool_directory=remote_tool_directory,
compute_job_directory=remote_job_directory,
)
# Pulsar handles ``create_tool_working_directory`` and
# ``include_work_dir_outputs`` details.
command_line = build_command(
self,
job_wrapper=job_wrapper,
container=container,
include_metadata=remote_metadata,
create_tool_working_directory=False,
include_work_dir_outputs=False,
remote_command_params=remote_command_params,
remote_job_directory=remote_job_directory,
)
except UnsupportedPulsarException:
log.exception("failure running job %d, unsupported Pulsar target", job_wrapper.job_id)
fail_or_resubmit = True
except PulsarClientTransportError:
log.exception("failure running job %d, Pulsar connection failed", job_wrapper.job_id)
fail_or_resubmit = True
except Exception:
log.exception("failure running job %d", job_wrapper.job_id)
fail_or_resubmit = True
# If we were unable to get a command line, there was problem
fail_or_resubmit = fail_or_resubmit or not command_line
if fail_or_resubmit:
job_state = self._job_state(job_wrapper.get_job(), job_wrapper)
self.work_queue.put((self.fail_job, job_state))
return command_line, client, remote_job_config, compute_environment, remote_container
def __prepare_input_files_locally(self, job_wrapper):
"""Run task splitting commands locally."""
prepare_input_files_cmds = getattr(job_wrapper, "prepare_input_files_cmds", None)
if prepare_input_files_cmds is not None:
for cmd in prepare_input_files_cmds: # run the commands to stage the input files
subprocess.check_call(cmd, shell=True)
job_wrapper.prepare_input_files_cmds = None # prevent them from being used in-line
def _populate_parameter_defaults(self, job_destination):
updated = False
params = job_destination.params
for key, value in self.destination_defaults.items():
if key in params:
if value is PARAMETER_SPECIFICATION_IGNORED:
log.warning(f"Pulsar runner in selected configuration ignores parameter {key}")
continue
# if self.runner_params.get( key, None ):
# # Let plugin define defaults for some parameters -
# # for instance that way jobs_directory can be
# # configured next to AMQP url (where it belongs).
# params[ key ] = self.runner_params[ key ]
# continue
if not value:
continue
if value is PARAMETER_SPECIFICATION_REQUIRED:
raise Exception(f"Pulsar destination does not define required parameter {key}")
elif value is not PARAMETER_SPECIFICATION_IGNORED:
params[key] = value
updated = True
return updated
def get_output_files(self, job_wrapper):
output_paths = job_wrapper.job_io.get_output_fnames()
return [str(o) for o in output_paths] # Force job_path from DatasetPath objects.
def get_input_files(self, job_wrapper):
input_paths = job_wrapper.job_io.get_input_paths()
return [str(i) for i in input_paths] # Force job_path from DatasetPath objects.
def get_client_from_wrapper(self, job_wrapper):
job_id = job_wrapper.job_id
if hasattr(job_wrapper, "task_id"):
job_id = f"{job_id}_{job_wrapper.task_id}"
params = job_wrapper.job_destination.params.copy()
user = job_wrapper.get_job().user
if user:
for key, value in params.items():
if value and isinstance(value, str):
params[key] = model.User.expand_user_properties(user, value)
env = getattr(job_wrapper.job_destination, "env", [])
return self.get_client(params, job_id, env)
def get_client_from_state(self, job_state):
job_destination_params = job_state.job_destination.params
job_id = job_state.job_wrapper.job_id # we want the Galaxy ID here, job_state.job_id is the external one.
return self.get_client(job_destination_params, job_id)
def get_client(self, job_destination_params, job_id, env=None):
# Cannot use url_for outside of web thread.
# files_endpoint = url_for( controller="job_files", job_id=encoded_job_id )
if env is None:
env = []
encoded_job_id = self.app.security.encode_id(job_id)
job_key = self.app.security.encode_id(job_id, kind="jobs_files")
endpoint_base = "%s/api/jobs/%s/files?job_key=%s"
if self.app.config.nginx_upload_job_files_path:
endpoint_base = "%s" + self.app.config.nginx_upload_job_files_path + "?job_id=%s&job_key=%s"
files_endpoint = endpoint_base % (self.galaxy_url, encoded_job_id, job_key)
secret = job_destination_params.get("job_secret_base", "jobs_token")
job_key = self.app.security.encode_id(job_id, kind=secret)
token_endpoint = f"{self.galaxy_url}/api/jobs/{encoded_job_id}/oidc-tokens?job_key={job_key}"
get_client_kwds = dict(
job_id=str(job_id), files_endpoint=files_endpoint, token_endpoint=token_endpoint, env=env
)
# Turn MutableDict into standard dict for pulsar consumption
job_destination_params = dict(job_destination_params.items())
return self.client_manager.get_client(job_destination_params, **get_client_kwds)
def finish_job(self, job_state: JobState):
assert isinstance(
job_state, AsynchronousJobState
), f"job_state type is '{type(job_state)}', expected AsynchronousJobState"
job_wrapper = job_state.job_wrapper
try:
client = self.get_client_from_state(job_state)
run_results = client.full_status()
remote_metadata_directory = run_results.get("metadata_directory", None)
tool_stdout = run_results.get("stdout", "")
tool_stderr = run_results.get("stderr", "")
job_stdout = run_results.get("job_stdout")
job_stderr = run_results.get("job_stderr")
exit_code = run_results.get("returncode")
pulsar_outputs = PulsarOutputs.from_status_response(run_results)
state = job_wrapper.get_state()
# Use Pulsar client code to transfer/copy files back
# and cleanup job if needed.
completed_normally = state not in [model.Job.states.ERROR, model.Job.states.DELETED]
if completed_normally and state == model.Job.states.STOPPED:
# Discard pulsar exit code (probably -9), we know the user stopped the job
log.debug("Setting exit code for stopped job {job_wrapper.job_id} to 0 (was {exit_code})")
exit_code = 0
cleanup_job = job_wrapper.cleanup_job
client_outputs = self.__client_outputs(client, job_wrapper)
finish_args = dict(
client=client,
job_completed_normally=completed_normally,
cleanup_job=cleanup_job,
client_outputs=client_outputs,
pulsar_outputs=pulsar_outputs,
)
failed = pulsar_finish_job(**finish_args)
if failed:
job_wrapper.fail(
"Failed to find or download one or more job outputs from remote server.", exception=True
)
except Exception:
self.fail_job(job_state, message=GENERIC_REMOTE_ERROR, exception=True)
log.exception("failure finishing job %d", job_wrapper.job_id)
return
if not PulsarJobRunner.__remote_metadata(client):
# we need an actual exit code file in the job working directory to detect job errors in the metadata script
with open(
os.path.join(job_wrapper.working_directory, f"galaxy_{job_wrapper.job_id}.ec"), "w"
) as exit_code_file:
exit_code_file.write(str(exit_code))
self._handle_metadata_externally(job_wrapper, resolve_requirements=True)
# Finish the job
try:
job_metrics_directory = os.path.join(job_wrapper.working_directory, "metadata")
# Following check is a hack for jobs started during 19.01 or earlier release
# and finishing with a 19.05 code base. Eliminate the hack in 19.09 or later
# along with hacks for legacy metadata compute strategy.
if not os.path.exists(job_metrics_directory) or not any(
"__instrument" in f for f in os.listdir(job_metrics_directory)
):
job_metrics_directory = job_wrapper.working_directory
job_wrapper.finish(
tool_stdout,
tool_stderr,
exit_code,
job_stdout=job_stdout,
job_stderr=job_stderr,
remote_metadata_directory=remote_metadata_directory,
job_metrics_directory=job_metrics_directory,
)
except Exception:
log.exception("Job wrapper finish method failed")
job_wrapper.fail("Unable to finish job", exception=True)
def check_pid(self, pid):
try:
os.kill(pid, 0)
return True
except OSError as e:
if e.errno == errno.ESRCH:
log.debug("check_pid(): PID %d is dead" % pid)
else:
log.warning(
"check_pid(): Got errno %s when attempting to check PID %d: %s"
% (errno.errorcode[e.errno], pid, e.strerror)
)
return False
def stop_job(self, job_wrapper):
job = job_wrapper.get_job()
if not job.job_runner_external_id:
return
# if our local job has JobExternalOutputMetadata associated, then our primary job has to have already finished
client = self.get_client(job.destination_params, job.job_runner_external_id)
job_ext_output_metadata = job.get_external_output_metadata()
if not PulsarJobRunner.__remote_metadata(client) and job_ext_output_metadata:
pid = job_ext_output_metadata[
0
].job_runner_external_pid # every JobExternalOutputMetadata has a pid set, we just need to take from one of them
if pid in [None, ""]:
log.warning(f"stop_job(): {job.id}: no PID in database for job, unable to stop")
return
pid = int(pid)
if not self.check_pid(pid):
log.warning("stop_job(): %s: PID %d was already dead or can't be signaled" % (job.id, pid))
return
for sig in [15, 9]:
try:
os.killpg(pid, sig)
except OSError as e:
log.warning(
"stop_job(): %s: Got errno %s when attempting to signal %d to PID %d: %s"
% (job.id, errno.errorcode[e.errno], sig, pid, e.strerror)
)
return # give up
sleep(2)
if not self.check_pid(pid):
log.debug("stop_job(): %s: PID %d successfully killed with signal %d" % (job.id, pid, sig))
return
else:
log.warning("stop_job(): %s: PID %d refuses to die after signaling TERM/KILL" % (job.id, pid))
else:
# Remote kill
pulsar_url = job.job_runner_name
job_id = job.job_runner_external_id
log.debug(f"Attempt remote Pulsar kill of job with url {pulsar_url} and id {job_id}")
client = self.get_client(job.destination_params, job_id)
client.kill()
def recover(self, job, job_wrapper):
"""Recover jobs stuck in the queued/running state when Galaxy started."""
job_state = self._job_state(job, job_wrapper)
job_wrapper.command_line = job.get_command_line()
state = job.get_state()
if state in [model.Job.states.RUNNING, model.Job.states.QUEUED, model.Job.states.STOPPED]:
log.debug(f"(Pulsar/{job.id}) is still in {state} state, adding to the Pulsar queue")
job_state.old_state = True
job_state.running = state == model.Job.states.RUNNING
self.monitor_queue.put(job_state)
def shutdown(self):
super().shutdown()
self.client_manager.shutdown()
if self.pulsar_app:
self.pulsar_app.shutdown()
def _job_state(self, job, job_wrapper):
raw_job_id = job.get_job_runner_external_id() or job_wrapper.job_id
job_state = AsynchronousJobState(
job_wrapper=job_wrapper, job_id=raw_job_id, job_destination=job_wrapper.job_destination
)
# TODO: Determine why this is set when using normal message queue updates
# but not CLI submitted MQ updates...
job_state.runner_url = job_wrapper.get_job_runner_url()
return job_state
def __client_outputs(self, client, job_wrapper):
metadata_directory = os.path.join(job_wrapper.working_directory, "metadata")
metadata_strategy = job_wrapper.get_destination_configuration("metadata_strategy", None)
tool = job_wrapper.tool
tool_provided_metadata_file_path = tool.provided_metadata_file
tool_provided_metadata_style = tool.provided_metadata_style
dynamic_outputs = None # use default
if metadata_strategy == "extended" and PulsarJobRunner.__remote_metadata(client):
# if Pulsar is doing remote metadata and the remote metadata is extended,
# we only need to recover the final model store.
dynamic_outputs = EXTENDED_METADATA_DYNAMIC_COLLECTION_PATTERN
output_files = []
work_dir_outputs = []
else:
# otherwise collect everything we might need
dynamic_outputs = DEFAULT_DYNAMIC_COLLECTION_PATTERN[:]
# grab discovered outputs...
dynamic_outputs.extend(job_wrapper.tool.output_discover_patterns)
# grab tool provided metadata (galaxy.json) also...
dynamic_outputs.append(re.escape(tool_provided_metadata_file_path))
output_files = self.get_output_files(job_wrapper)
work_dir_outputs = self.get_work_dir_outputs(job_wrapper)
dynamic_file_sources = [
{
"path": tool_provided_metadata_file_path,
"type": "galaxy" if tool_provided_metadata_style == "default" else "legacy_galaxy",
}
]
client_outputs = ClientOutputs(
working_directory=job_wrapper.tool_working_directory,
metadata_directory=metadata_directory,
work_dir_outputs=work_dir_outputs,
output_files=output_files,
version_file=job_wrapper.get_version_string_path(),
dynamic_outputs=dynamic_outputs,
dynamic_file_sources=dynamic_file_sources,
)
return client_outputs
@staticmethod
def pulsar_version(remote_job_config):
pulsar_version = Version(remote_job_config.get("pulsar_version", "0.6.0"))
return pulsar_version
@staticmethod
def check_job_config(remote_job_config, check_features=None):
check_features = check_features or {}
# 0.6.0 was newest Pulsar version that did not report it's version.
pulsar_version = PulsarJobRunner.pulsar_version(remote_job_config)
needed_version = Version("0.0.0")
log.info(f"pulsar_version is {pulsar_version}")
for feature, needed in list(check_features.items()) + [("_default_", True)]:
if not needed:
continue
if pulsar_version < MINIMUM_PULSAR_VERSIONS[feature]:
needed_version = max(needed_version, MINIMUM_PULSAR_VERSIONS[feature])
if pulsar_version < needed_version:
raise UnsupportedPulsarException(needed_version)
@staticmethod
def __dependencies_description(pulsar_client, job_wrapper):
dependency_resolution = PulsarJobRunner.__dependency_resolution(pulsar_client)
remote_dependency_resolution = dependency_resolution == "remote"
if not remote_dependency_resolution:
return None
requirements = job_wrapper.tool.requirements
installed_tool_dependencies = job_wrapper.tool.installed_tool_dependencies
return dependencies.DependenciesDescription(
requirements=requirements,
installed_tool_dependencies=installed_tool_dependencies,
)
@staticmethod
def __dependency_resolution(pulsar_client):
dependency_resolution = pulsar_client.destination_params.get("dependency_resolution", "remote")
if dependency_resolution not in ["none", "local", "remote"]:
raise Exception(f"Unknown dependency_resolution value encountered {dependency_resolution}")
return dependency_resolution
@staticmethod
def __remote_metadata(pulsar_client):
remote_metadata = string_as_bool_or_none(pulsar_client.destination_params.get("remote_metadata", False))
return remote_metadata
@staticmethod
def __remote_container_handling(pulsar_client):
remote_container_handling = string_as_bool_or_none(
pulsar_client.destination_params.get("remote_container_handling", False)
)
return remote_container_handling
@staticmethod
def __use_remote_datatypes_conf(pulsar_client):
"""Use remote metadata datatypes instead of Galaxy's.
When setting remote metadata, use integrated datatypes from this
Galaxy instance or use the datatypes config configured via the remote
Pulsar.
Both options are broken in different ways for same reason - datatypes
may not match. One can push the local datatypes config to the remote
server - but there is no guarentee these datatypes will be defined
there. Alternatively, one can use the remote datatype config - but
there is no guarentee that it will contain all the datatypes available
to this Galaxy.
"""
use_remote_datatypes = string_as_bool_or_none(
pulsar_client.destination_params.get("use_remote_datatypes", False)
)
return use_remote_datatypes
@staticmethod
def __rewrite_parameters(pulsar_client):
return string_as_bool_or_none(pulsar_client.destination_params.get("rewrite_parameters", False)) or False
def __build_metadata_configuration(
self,
client,
job_wrapper,
remote_metadata,
remote_job_config,
compute_environment: Optional["PulsarComputeEnvironment"] = None,
):
metadata_kwds: Dict[str, Any] = {}
if remote_metadata:
working_directory = remote_job_config["working_directory"]
metadata_directory = remote_job_config["metadata_directory"]
# For metadata calculation, we need to build a list of of output
# file objects with real path indicating location on Galaxy server
# and false path indicating location on compute server. Since the
# Pulsar disables from_work_dir copying as part of the job command
# line we need to take the list of output locations on the Pulsar
# server (produced by job_wrapper.job_io.get_output_fnames() and for
# each work_dir output substitute the effective path on the Pulsar
# server relative to the remote working directory as the
# false_path to send the metadata command generation module.
work_dir_outputs = self.get_work_dir_outputs(job_wrapper, tool_working_directory=working_directory)
outputs = job_wrapper.job_io.get_output_fnames()
# copy fixes 'test/integration/test_pulsar_embedded_remote_metadata.py::test_tools[job_properties]'
real_path_to_output = {o.real_path: copy.copy(o) for o in outputs}
rewritten_outputs = []
for pulsar_workdir_path, real_path in work_dir_outputs:
work_dir_output = real_path_to_output.pop(real_path, None)
if work_dir_output:
work_dir_output.false_path = pulsar_workdir_path
rewritten_outputs.append(work_dir_output)
for output in real_path_to_output.values():
if compute_environment:
output.false_path = compute_environment.path_mapper.remote_output_path_rewrite(str(output))
rewritten_outputs.append(output)
metadata_kwds["output_fnames"] = rewritten_outputs
remote_system_properties = remote_job_config.get("system_properties", {})
remote_galaxy_home = remote_system_properties.get("galaxy_home")
if not job_wrapper.use_metadata_binary:
if not remote_galaxy_home:
raise Exception(NO_REMOTE_GALAXY_FOR_METADATA_MESSAGE)
metadata_kwds["exec_dir"] = remote_galaxy_home
metadata_kwds["compute_tmp_dir"] = metadata_directory
metadata_kwds["config_root"] = remote_galaxy_home
default_config_file = os.path.join(remote_galaxy_home, "config/galaxy.ini")
metadata_kwds["config_file"] = remote_system_properties.get("galaxy_config_file", default_config_file)
metadata_kwds["dataset_files_path"] = remote_system_properties.get("galaxy_dataset_files_path", None)
if PulsarJobRunner.__use_remote_datatypes_conf(client):
remote_datatypes_config = remote_system_properties.get("galaxy_datatypes_config_file")
if not remote_datatypes_config:
log.warning(NO_REMOTE_DATATYPES_CONFIG)
if not remote_galaxy_home:
raise Exception(NO_REMOTE_GALAXY_FOR_METADATA_MESSAGE)
remote_datatypes_config = os.path.join(remote_galaxy_home, "datatypes_conf.xml")
metadata_kwds["datatypes_config"] = remote_datatypes_config
else:
datatypes_config = os.path.join(job_wrapper.working_directory, "registry.xml")
self.app.datatypes_registry.to_xml_file(path=datatypes_config)
# Ensure this file gets pushed out to the remote config dir.
job_wrapper.extra_filenames.append(datatypes_config)
metadata_kwds["datatypes_config"] = datatypes_config
return metadata_kwds
def __async_update(self, full_status):
galaxy_job_id = None
remote_job_id = None
try:
remote_job_id = full_status["job_id"]
if len(remote_job_id) == 32:
# It is a UUID - assign_ids = uuid in destination params...
sa_session = self.app.model.session
galaxy_job_id = (
sa_session.query(model.Job).filter(model.Job.job_runner_external_id == remote_job_id).one().id
)
else:
galaxy_job_id = remote_job_id
job, job_wrapper = self.app.job_manager.job_handler.job_queue.job_pair_for_id(galaxy_job_id)
job_state = self._job_state(job, job_wrapper)
self._update_job_state_for_status(job_state, full_status["status"], full_status=full_status)
except Exception:
log.exception(f"Failed to update Pulsar job status for job_id ({galaxy_job_id}/{remote_job_id})")
raise
# Nothing else to do? - Attempt to fail the job?
[docs]class PulsarLegacyJobRunner(PulsarJobRunner):
"""Flavor of Pulsar job runner mimicking behavior of old LWR runner."""
destination_defaults = dict(
rewrite_parameters="false",
dependency_resolution="local",
)
[docs]class PulsarMQJobRunner(PulsarJobRunner):
"""Flavor of Pulsar job runner with sensible defaults for message queue communication."""
use_mq = True
poll = False
destination_defaults = dict(
default_file_action="remote_transfer",
rewrite_parameters="true",
dependency_resolution="remote",
jobs_directory=PARAMETER_SPECIFICATION_REQUIRED,
url=PARAMETER_SPECIFICATION_IGNORED,
private_token=PARAMETER_SPECIFICATION_IGNORED,
)
DEFAULT_PULSAR_CONTAINER = "galaxy/pulsar-pod-staging:0.15.0.2"
COEXECUTION_DESTENTATION_DEFAULTS = {
"default_file_action": "remote_transfer",
"rewrite_parameters": "true",
"jobs_directory": "/pulsar_staging",
"pulsar_container_image": DEFAULT_PULSAR_CONTAINER,
"remote_container_handling": True,
"url": PARAMETER_SPECIFICATION_IGNORED,
"private_token": PARAMETER_SPECIFICATION_IGNORED,
}
class PulsarCoexecutionJobRunner(PulsarMQJobRunner):
destination_defaults = COEXECUTION_DESTENTATION_DEFAULTS
def _populate_parameter_defaults(self, job_destination):
super()._populate_parameter_defaults(job_destination)
params = job_destination.params
# Set some sensible defaults for Pulsar application that runs in staging container.
if "pulsar_app_config" not in params:
params["pulsar_app_config"] = {}
pulsar_app_config = params["pulsar_app_config"]
if "staging_directory" not in pulsar_app_config:
# coexecution always uses a fixed path for staging directory
pulsar_app_config["staging_directory"] = params.get("jobs_directory")
KUBERNETES_DESTINATION_DEFAULTS: Dict[str, Any] = {"k8s_enabled": True, **COEXECUTION_DESTENTATION_DEFAULTS}
class PulsarKubernetesJobRunner(PulsarCoexecutionJobRunner):
destination_defaults = KUBERNETES_DESTINATION_DEFAULTS
poll = True # Poll so we can check API for pod IP for ITs.
TES_DESTENTATION_DEFAULTS: Dict[str, Any] = {
"tes_url": PARAMETER_SPECIFICATION_REQUIRED,
**COEXECUTION_DESTENTATION_DEFAULTS,
}
class PulsarTesJobRunner(PulsarCoexecutionJobRunner):
destination_defaults = TES_DESTENTATION_DEFAULTS
[docs]class PulsarRESTJobRunner(PulsarJobRunner):
"""Flavor of Pulsar job runner with sensible defaults for RESTful usage."""
destination_defaults = dict(
default_file_action="transfer",
rewrite_parameters="true",
dependency_resolution="remote",
url=PARAMETER_SPECIFICATION_REQUIRED,
)
[docs]class PulsarEmbeddedJobRunner(PulsarJobRunner):
"""Flavor of Puslar job runner that runs Pulsar's server code directly within Galaxy.
This is an appropriate job runner for when the desire is to use Pulsar staging
but their is not need to run a remote service.
"""
destination_defaults = dict(
default_file_action="copy",
rewrite_parameters="true",
dependency_resolution="remote",
)
default_build_pulsar_app = True
class PulsarComputeEnvironment(ComputeEnvironment):
def __init__(self, pulsar_client, job_wrapper, remote_job_config):
self.pulsar_client = pulsar_client
self.job_wrapper = job_wrapper
self.local_path_config = job_wrapper.default_compute_environment()
self.path_rewrites_unstructured = {}
self.path_rewrites_input_extra = {}
self.path_rewrites_input_metadata = {}
# job_wrapper.prepare is going to expunge the job backing the following
# computations, so precalculate these paths.
self.path_mapper = PathMapper(pulsar_client, remote_job_config, self.local_path_config.working_directory())
self._config_directory = remote_job_config["configs_directory"]
self._working_directory = remote_job_config["working_directory"]
self._sep = remote_job_config["system_properties"]["separator"]
self._tool_dir = remote_job_config["tools_directory"]
self._tmp_dir = remote_job_config.get("tmp_dir")
self._shared_home_dir = remote_job_config.get("shared_home_dir")
version_path = self.local_path_config.version_path()
new_version_path = self.path_mapper.remote_version_path_rewrite(version_path)
if new_version_path:
version_path = new_version_path
self._version_path = version_path
def output_names(self):
# Maybe this should use the path mapper, but the path mapper just uses basenames
return self.job_wrapper.job_io.get_output_basenames()
def input_path_rewrite(self, dataset):
local_input_path_rewrite = self.local_path_config.input_path_rewrite(dataset)
if local_input_path_rewrite is not None:
local_input_path = local_input_path_rewrite
else:
local_input_path = dataset.file_name
remote_path = self.path_mapper.remote_input_path_rewrite(local_input_path)
return remote_path
def output_path_rewrite(self, dataset):
local_output_path_rewrite = self.local_path_config.output_path_rewrite(dataset)
if local_output_path_rewrite is not None:
local_output_path = local_output_path_rewrite
else:
local_output_path = dataset.file_name
remote_path = self.path_mapper.remote_output_path_rewrite(local_output_path)
return remote_path
def input_extra_files_rewrite(self, dataset):
input_path_rewrite = self.input_path_rewrite(dataset)
remote_extra_files_path_rewrite = dataset_path_to_extra_path(input_path_rewrite)
self.path_rewrites_input_extra[dataset.extra_files_path] = remote_extra_files_path_rewrite
return remote_extra_files_path_rewrite
def output_extra_files_rewrite(self, dataset):
output_path_rewrite = self.output_path_rewrite(dataset)
remote_extra_files_path_rewrite = dataset_path_to_extra_path(output_path_rewrite)
return remote_extra_files_path_rewrite
def input_metadata_rewrite(self, dataset, metadata_val):
# May technically be incorrect to not pass through local_path_config.input_metadata_rewrite
# first but that adds untested logic that wouln't ever be used.
remote_input_path = self.path_mapper.remote_input_path_rewrite(
metadata_val, client_input_path_type=CLIENT_INPUT_PATH_TYPES.INPUT_METADATA_PATH
)
if remote_input_path:
log.info(f"input_metadata_rewrite is {remote_input_path} from {metadata_val}")
self.path_rewrites_input_metadata[metadata_val] = remote_input_path
return remote_input_path
# No rewrite...
return None
def unstructured_path_rewrite(self, parameter_value):
path_rewrites_unstructured = self.path_rewrites_unstructured
if parameter_value in path_rewrites_unstructured:
# Path previously mapped, use previous mapping.
return path_rewrites_unstructured[parameter_value]
rewrite, new_unstructured_path_rewrites = self.path_mapper.check_for_arbitrary_rewrite(parameter_value)
if rewrite:
path_rewrites_unstructured.update(new_unstructured_path_rewrites)
return rewrite
else:
# Did not need to rewrite, use original path or value.
return None
def working_directory(self):
return self._working_directory
def env_config_directory(self):
return self.config_directory()
def config_directory(self):
return self._config_directory
def new_file_path(self):
return self.working_directory() # Problems with doing this?
def sep(self):
return self._sep
def version_path(self):
return self._version_path
def tool_directory(self):
return self._tool_dir
def home_directory(self):
return self._target_to_directory(self.job_wrapper.home_target)
def tmp_directory(self):
return self._target_to_directory(self.job_wrapper.tmp_target)
def _target_to_directory(self, target):
tmp_dir = self._tmp_dir
if target is None or (target == "job_tmp_if_explicit" and tmp_dir is None):
return None
elif target in ["job_tmp", "job_tmp_if_explicit"]:
return "$_GALAXY_JOB_TMP_DIR"
elif target == "shared_home":
return self._shared_home_dir
elif target == "job_home":
return "$_GALAXY_JOB_HOME_DIR"
elif target == "pwd":
os.path.join(self.working_directory(), "working")
else:
raise Exception(f"Unknown target type [{target}]")
def galaxy_url(self):
return self.job_wrapper.get_destination_configuration("galaxy_infrastructure_url")
def get_file_sources_dict(self):
return self.job_wrapper.job_io.file_sources_dict
class UnsupportedPulsarException(Exception):
def __init__(self, needed):
super().__init__(UPGRADE_PULSAR_ERROR % needed)