X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/b0e2fe9d0a18d5e809bc8c0d3382e0e023cb949c..44c95f99098fa6c6acbfa82d4b6cbc6015eb6e39:/sdk/python/arvados/__init__.py diff --git a/sdk/python/arvados/__init__.py b/sdk/python/arvados/__init__.py index 19a7ad8439..06a4f07a79 100644 --- a/sdk/python/arvados/__init__.py +++ b/sdk/python/arvados/__init__.py @@ -1,30 +1,43 @@ -import gflags -import httplib +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + +from __future__ import print_function +from __future__ import absolute_import +from future import standard_library +standard_library.install_aliases() +from builtins import object +import bz2 +import fcntl +import hashlib +import http.client import httplib2 +import json import logging import os import pprint -import sys -import types -import subprocess -import json -import UserDict import re -import hashlib import string -import bz2 -import zlib -import fcntl -import time +import subprocess +import sys import threading +import time +import types +import zlib + +if sys.version_info >= (3, 0): + from collections import UserDict +else: + from UserDict import UserDict -from api import * -from collection import * -from keep import * -from stream import * -from arvfile import * -import errors -import util +from .api import api, api_from_config, http_cache +from .collection import CollectionReader, CollectionWriter, ResumableCollectionWriter +from arvados.keep import * +from arvados.stream import * +from .arvfile import StreamFileReader +from .retry import RetryLoop +import arvados.errors as errors +import arvados.util as util # Set up Arvados logging based on the user's configuration. # All Arvados code should log under the arvados hierarchy. @@ -37,36 +50,60 @@ logger.addHandler(log_handler) logger.setLevel(logging.DEBUG if config.get('ARVADOS_DEBUG') else logging.WARNING) -def task_set_output(self,s): - api('v1').job_tasks().update(uuid=self['uuid'], - body={ - 'output':s, - 'success':True, - 'progress':1.0 - }).execute() +def task_set_output(self, s, num_retries=5): + for tries_left in RetryLoop(num_retries=num_retries, backoff_start=0): + try: + return api('v1').job_tasks().update( + uuid=self['uuid'], + body={ + 'output':s, + 'success':True, + 'progress':1.0 + }).execute() + except errors.ApiError as error: + if retry.check_http_response_success(error.resp.status) is None and tries_left > 0: + logger.debug("task_set_output: job_tasks().update() raised {}, retrying with {} tries left".format(repr(error),tries_left)) + else: + raise _current_task = None -def current_task(): +def current_task(num_retries=5): global _current_task if _current_task: return _current_task - t = api('v1').job_tasks().get(uuid=os.environ['TASK_UUID']).execute() - t = UserDict.UserDict(t) - t.set_output = types.MethodType(task_set_output, t) - t.tmpdir = os.environ['TASK_WORK'] - _current_task = t - return t + + for tries_left in RetryLoop(num_retries=num_retries, backoff_start=2): + try: + task = api('v1').job_tasks().get(uuid=os.environ['TASK_UUID']).execute() + task = UserDict(task) + task.set_output = types.MethodType(task_set_output, task) + task.tmpdir = os.environ['TASK_WORK'] + _current_task = task + return task + except errors.ApiError as error: + if retry.check_http_response_success(error.resp.status) is None and tries_left > 0: + logger.debug("current_task: job_tasks().get() raised {}, retrying with {} tries left".format(repr(error),tries_left)) + else: + raise _current_job = None -def current_job(): +def current_job(num_retries=5): global _current_job if _current_job: return _current_job - t = api('v1').jobs().get(uuid=os.environ['JOB_UUID']).execute() - t = UserDict.UserDict(t) - t.tmpdir = os.environ['JOB_WORK'] - _current_job = t - return t + + for tries_left in RetryLoop(num_retries=num_retries, backoff_start=2): + try: + job = api('v1').jobs().get(uuid=os.environ['JOB_UUID']).execute() + job = UserDict(job) + job.tmpdir = os.environ['JOB_WORK'] + _current_job = job + return job + except errors.ApiError as error: + if retry.check_http_response_success(error.resp.status) is None and tries_left > 0: + logger.debug("current_job: jobs().get() raised {}, retrying with {} tries left".format(repr(error),tries_left)) + else: + raise def getjobparam(*args): return current_job()['script_parameters'].get(*args) @@ -79,15 +116,19 @@ def get_task_param_mount(*args): class JobTask(object): def __init__(self, parameters=dict(), runtime_constraints=dict()): - print "init jobtask %s %s" % (parameters, runtime_constraints) + print("init jobtask %s %s" % (parameters, runtime_constraints)) -class job_setup: +class job_setup(object): @staticmethod - def one_task_per_input_file(if_sequence=0, and_end_task=True, input_as_path=False): + def one_task_per_input_file(if_sequence=0, and_end_task=True, input_as_path=False, api_client=None): if if_sequence != current_task()['sequence']: return + + if not api_client: + api_client = api('v1') + job_input = current_job()['script_parameters']['input'] - cr = CollectionReader(job_input) + cr = CollectionReader(job_input, api_client=api_client) cr.normalize() for s in cr.all_streams(): for f in s.all_files(): @@ -103,9 +144,9 @@ class job_setup: 'input':task_input } } - api('v1').job_tasks().create(body=new_task_attrs).execute() + api_client.job_tasks().create(body=new_task_attrs).execute() if and_end_task: - api('v1').job_tasks().update(uuid=current_task()['uuid'], + api_client.job_tasks().update(uuid=current_task()['uuid'], body={'success':True} ).execute() exit(0)