def download_live_logs(args):
logfetch_base.log(colored('Finding current live log files', 'cyan') + '\n', args, False)
tasks = tasks_to_check(args)
async_requests = []
all_logs = []
callbacks.progress = 0
tasks_check_progress = 0
tasks_check_goal = len(tasks)
for task in tasks:
metadata = files_json(args, task)
if 'slaveHostname' in metadata:
uri = DOWNLOAD_FILE_FORMAT.format(metadata['slaveHostname'])
for log_file in base_directory_files(args, task, metadata):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (args.logtype and logfetch_base.log_matches(log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest('GET',uri ,
callback=callbacks.generate_callback(uri, args.dest, logfile_name, args.chunk_size, args.verbose, args.silent),
params={'path' : '{0}/{1}/{2}'.format(metadata['fullPathToRoot'], metadata['currentDirectory'], log_file)},
headers=args.headers
)
)
all_logs.append('{0}/{1}'.format(args.dest, logfile_name))
elif args.logtype:
logfetch_base.log(colored('Excluding log {0}, doesn\'t match {1}'.format(log_file, args.logtype), 'magenta') + '\n', args, True)
for log_file in logs_folder_files(args, task):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (args.logtype and logfetch_base.log_matches(log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest('GET',uri ,
callback=callbacks.generate_callback(uri, args.dest, logfile_name, args.chunk_size, args.verbose, args.silent),
params={'path' : '{0}/{1}/logs/{2}'.format(metadata['fullPathToRoot'], metadata['currentDirectory'], log_file)},
headers=args.headers
)
)
all_logs.append('{0}/{1}'.format(args.dest, logfile_name))
elif args.logtype:
logfetch_base.log(colored('Excluding log {0}, doesn\'t match {1}'.format(log_file, args.logtype), 'magenta') + '\n', args, True)
tasks_check_progress += 1
logfetch_base.update_progress_bar(tasks_check_progress, tasks_check_goal, 'Log Finder', args.silent)
if async_requests:
logfetch_base.log(colored('\nStarting {0} live logs downloads\n'.format(len(async_requests)), 'cyan'), args, False)
callbacks.goal = len(async_requests)
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
return all_logs
def download_s3_logs(args):
if not args.silent:
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
callbacks.progress = 0
logs = logs_for_all_requests(args)
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(args, int(str(log_file['lastModified'])[0:-3])):
if not args.logtype or log_matches(args, filename):
logfetch_base.log(colored('Including log {0}'.format(filename), 'blue') + '\n', args, True)
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'], callback=callbacks.generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size, args.verbose, args.silent), headers=args.headers)
)
else:
logfetch_base.log(colored('Log already downloaded {0}'.format(filename), 'blue') + '\n', args, True)
all_logs.append('{0}/{1}'.format(args.dest, filename))
else:
logfetch_base.log(colored('Excluding {0} log does not match logtype argument {1}'.format(filename, args.logtype), 'magenta') + '\n', args, True)
else:
logfetch_base.log(colored('Excluding {0}, not in date range'.format(filename), 'magenta') + '\n', args, True)
if async_requests:
logfetch_base.log(colored('Starting {0} S3 Downloads with {1} parallel fetches\n'.format(len(async_requests), args.num_parallel_fetches), 'cyan'), args, False)
callbacks.goal = len(async_requests)
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
else:
logfetch_base.log(colored('No S3 logs to download\n', 'cyan'), args, False)
logfetch_base.log(colored('All S3 logs up to date\n', 'cyan'), args, False)
all_logs = modify_download_list(all_logs)
return all_logs
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'cyan') + '\n')
logs = logs_for_all_requests(args)
async_requests = []
zipped_files = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
if logfetch_base.is_in_date_range(
args, int(str(log_file['lastModified'])[0:-3])):
if not args.logtype or log_matches(args, filename):
if not already_downloaded(args.dest, filename):
async_requests.append(
grequests.AsyncRequest('GET',
log_file['getUrl'],
callback=generate_callback(
log_file['getUrl'],
args.dest, filename,
args.chunk_size,
args.verbose),
headers=args.headers))
else:
if args.verbose:
sys.stderr.write(
colored(
'Log already downloaded {0}'.format(filename),
'magenta') + '\n')
all_logs.append('{0}/{1}'.format(
args.dest, filename.replace('.gz', '.log')))
zipped_files.append('{0}/{1}'.format(args.dest, filename))
else:
if args.verbose:
sys.stderr.write(
colored(
'Excluding {0} log does not match logtype argument {1}'
.format(filename, args.logtype), 'magenta') + '\n')
else:
if args.verbose:
sys.stderr.write(
colored(
'Excluding {0}, not in date range'.format(filename),
'magenta') + '\n')
if async_requests:
sys.stderr.write(
colored(
'Starting S3 Downloads with {0} parallel fetches'.format(
args.num_parallel_fetches), 'cyan'))
grequests.map(async_requests,
stream=True,
size=args.num_parallel_fetches)
else:
sys.stderr.write(colored('No S3 logs to download', 'cyan'))
sys.stderr.write(colored('\nUnpacking S3 logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'cyan') + '\n')
return all_logs
def do(self, action, name, args, asynchronous=False):
"Perform action"
if args.pop('__meta', None):
return (dict(api='/api/%s/%s' % (self.version, name),
method=action.upper(),
args=args))
self._ensureKey()
url = '%s/api/%s/%s' % (self.host, self.version, name)
self._debug('> %s %s' % (action.upper(), url))
kw = {}
if action == 'get':
body = kw['params'] = args
else:
body = kw['data'] = json.dumps(args, indent=1, cls=JSONEncoder)
if self.timeout: kw['timeout'] = self.timeout
headers = {'content-type': 'application/json'}
headers.update(self._requestAuthHeaders)
for k, v in list(self._requestAuthHeaders.items()):
self._debug('>> %s: %s' % (k, v))
if body:
self._debug("\n%s\n" % body)
if self.xml:
headers['accept'] = 'application/xml'
kw['verify'] = self.verifySSL
headers.update(self.headers)
if asynchronous:
import grequests
return grequests.AsyncRequest(action, url, headers=headers, **kw)
try:
r = self.session.request(action, url, headers=headers, **kw)
except requests.ConnectionError as e:
raise BeaconAPIException(
code=500,
message="Could not connect to server (%s): %s" % (url, e))
self._debug('<< %s %s' % (r.status_code, r.reason))
if 'x-typehint' in r.headers:
self._debug('< x-typehint: %s' % r.headers['x-typehint'])
if r.status_code != 200:
raise BeaconAPIException(code=r.status_code,
message=r.reason,
body=r.content)
if r.headers['content-type'] == 'application/json':
return r.json()
return r.content
def main(parser, args):
if args.taskId:
singularity_path = "task/%s" % args.taskId
elif args.deployId:
if not args.requestId:
exit(parser, "RequestId requires DeployId")
singularity_path = "request/%s/deploy/%s" % (args.requestId,
args.deployId)
elif args.requestId:
singularity_path = "request/%s" % args.requestId
else:
exit(parser,
"Specify one of taskId, requestId and deployId, or requestId")
if not args.singularity_uri_base:
exit(parser, "Specify a base uri for Singularity")
uri_prefix = ""
if not args.singularity_uri_base.startswith(("http://", "https://")):
uri_prefix = "http://"
singularity_uri = "%s%s/logs/%s" % (uri_prefix, args.singularity_uri_base,
singularity_path)
print "fetching log metadata from %s" % singularity_uri
singularity_response = requests.get(singularity_uri)
if singularity_response.status_code < 199 or singularity_response.status_code > 299:
exit(
None, "Singularity responded with an invalid status code (%s)" %
singularity_response.status_code)
s_json = singularity_response.json()
print "found %s log files" % len(s_json)
async_requests = [
grequests.AsyncRequest('GET',
log_file['getUrl'],
callback=generate_callback(
log_file['getUrl'], args.dest,
log_file['key'][log_file['key'].rfind('/') +
1:], args.chunk_size))
for log_file in s_json
]
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
def download_s3_logs(args):
sys.stderr.write(colored('Checking for S3 log files', 'blue') + '\n')
logs = get_json_response(singularity_s3logs_uri(args))
async_requests = []
all_logs = []
for log_file in logs:
filename = log_file['key'].rsplit("/", 1)[1]
full_log_path = '{0}/{1}'.format(args.dest, filename.replace('.gz', '.log'))
full_gz_path = '{0}/{1}'.format(args.dest, filename)
if in_date_range(args, filename):
if not (os.path.isfile(full_log_path) or os.path.isfile(full_gz_path)):
async_requests.append(
grequests.AsyncRequest('GET', log_file['getUrl'],
callback=generate_callback(log_file['getUrl'], args.dest, filename, args.chunk_size)
)
)
all_logs.append('{0}/{1}'.format(args.dest, filename.replace('.gz', '.log')))
grequests.map(async_requests, stream=True, size=args.num_parallel_fetches)
zipped_files = ['{0}/{1}'.format(args.dest, log_file['key'].rsplit("/", 1)[1]) for log_file in logs]
logfetch_base.unpack_logs(zipped_files)
sys.stderr.write(colored('All S3 logs up to date', 'blue') + '\n')
return all_logs
def download_live_logs(args):
tasks = tasks_to_check(args)
async_requests = []
zipped_files = []
all_logs = []
sys.stderr.write(colored('Finding current live log files', 'cyan') + '\n')
for task in tasks:
metadata = files_json(args, task)
if 'slaveHostname' in metadata:
uri = DOWNLOAD_FILE_FORMAT.format(metadata['slaveHostname'])
if args.verbose:
sys.stderr.write(
colored(
'Finding logs in base directory on {0}'.format(
metadata['slaveHostname']), 'magenta') + '\n')
for log_file in base_directory_files(args, task, metadata):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (
args.logtype and logfetch_base.log_matches(
log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest(
'GET',
uri,
callback=generate_callback(
uri, args.dest, logfile_name,
args.chunk_size, args.verbose),
params={
'path':
'{0}/{1}/{2}'.format(
metadata['fullPathToRoot'],
metadata['currentDirectory'], log_file)
},
headers=args.headers))
if logfile_name.endswith('.gz'):
zipped_files.append('{0}/{1}'.format(
args.dest, logfile_name))
else:
all_logs.append('{0}/{1}'.format(
args.dest, logfile_name.replace('.gz', '.log')))
elif args.logtype and args.verbose:
sys.stderr.write(
colored(
'Excluding log {0}, doesn\'t match {1}'.format(
log_file, args.logtype), 'magenta') + '\n')
if args.verbose:
sys.stderr.write(
colored(
'Finding logs in logs directory on {0}'.format(
metadata['slaveHostname']), 'magenta') + '\n')
for log_file in logs_folder_files(args, task):
logfile_name = '{0}-{1}'.format(task, log_file)
if not args.logtype or (
args.logtype and logfetch_base.log_matches(
log_file, args.logtype.replace('logs/', ''))):
if should_download(args, logfile_name, task):
async_requests.append(
grequests.AsyncRequest(
'GET',
uri,
callback=generate_callback(
uri, args.dest, logfile_name,
args.chunk_size, args.verbose),
params={
'path':
'{0}/{1}/logs/{2}'.format(
metadata['fullPathToRoot'],
metadata['currentDirectory'], log_file)
},
headers=args.headers))
if logfile_name.endswith('.gz'):
zipped_files.append('{0}/{1}'.format(
args.dest, logfile_name))
else:
all_logs.append('{0}/{1}'.format(
args.dest, logfile_name.replace('.gz', '.log')))
elif args.logtype and args.verbose:
sys.stderr.write(
colored(
'Excluding log {0}, doesn\'t match {1}'.format(
log_file, args.logtype), 'magenta') + '\n')
if async_requests:
sys.stderr.write(colored('Starting live logs downloads\n', 'cyan'))
grequests.map(async_requests,
stream=True,
size=args.num_parallel_fetches)
if zipped_files:
sys.stderr.write(colored('\nUnpacking logs\n', 'cyan'))
all_logs = all_logs + logfetch_base.unpack_logs(args, zipped_files)
return all_logs
# -*- coding: utf-8 -*-
"""
grequests 是 gevent + requests
"""
from urls import url_list
import grequests
from sfm.timer import DateTimeTimer as Timer
with Timer(title="use grequests"):
req_list = [
grequests.AsyncRequest(method="GET", url=url)
for url in url_list
]
res_list = grequests.map(req_list)
results = [len(res.text) for res in res_list]
print(results)