def _get_all_pages(self):
got = 0
last_id = 0
filters = [
['object_uuid', '=', self.job_uuid],
['event_type', '=', 'stderr']]
try:
while True:
page = arvados.api().logs().index(
limit=1000,
order=['id asc'],
filters=filters + [['id','>',str(last_id)]],
select=['id', 'properties'],
).execute(num_retries=2)
got += len(page['items'])
logger.debug(
'%s: received %d of %d log events',
self.job_uuid, got,
got + page['items_available'] - len(page['items']))
for i in page['items']:
for line in i['properties']['text'].split('\n'):
self._queue.put(line+'\n')
last_id = i['id']
if (len(page['items']) == 0 or
len(page['items']) >= page['items_available']):
break
finally:
self._queue.put(self.EOF)
def _get_all_pages(self):
got = 0
last_id = 0
filters = [['object_uuid', '=', self.job_uuid],
['event_type', '=', 'stderr']]
try:
while True:
page = arvados.api().logs().index(
limit=1000,
order=['id asc'],
filters=filters + [['id', '>', str(last_id)]],
select=['id', 'properties'],
).execute(num_retries=2)
got += len(page['items'])
logger.debug(
'%s: received %d of %d log events', self.job_uuid, got,
got + page['items_available'] - len(page['items']))
for i in page['items']:
for line in i['properties']['text'].split('\n'):
self._queue.put(line + '\n')
last_id = i['id']
if (len(page['items']) == 0
or len(page['items']) >= page['items_available']):
break
finally:
self._queue.put(self.EOF)
def __init__(self, collection_id):
logger.debug('load collection %s', collection_id)
collection = arvados.collection.CollectionReader(collection_id)
filenames = [filename for filename in collection]
if len(filenames) != 1:
raise ValueError(
"collection {} has {} files; need exactly one".format(
collection_id, len(filenames)))
self._reader = collection.open(filenames[0])
self._label = "{}/{}".format(collection_id, filenames[0])
def __init__(self, collection_id):
logger.debug('load collection %s', collection_id)
collection = arvados.collection.CollectionReader(collection_id)
filenames = [filename for filename in collection]
if len(filenames) != 1:
raise ValueError(
"collection {} has {} files; need exactly one".format(
collection_id, len(filenames)))
self._reader = collection.open(filenames[0])
self._label = "{}/{}".format(collection_id, filenames[0])
def __init__(self, logdata, label=None, include_crunchstat_summary=False):
self._logdata = logdata
self.label = label
self._include_crunchstat_summary = include_crunchstat_summary
self.seq_to_uuid = {}
self.tasks = collections.defaultdict(Task)
# We won't bother recommending new runtime constraints if the
# constraints given when running the job are known to us and
# are already suitable. If applicable, the subclass
# constructor will overwrite this with something useful.
self.existing_constraints = {}
logger.debug("%s: logdata %s", self.label, logdata)
def __init__(self, logdata, label=None, include_crunchstat_summary=False):
self._logdata = logdata
self.label = label
self._include_crunchstat_summary = include_crunchstat_summary
self.seq_to_uuid = {}
self.tasks = collections.defaultdict(Task)
# We won't bother recommending new runtime constraints if the
# constraints given when running the job are known to us and
# are already suitable. If applicable, the subclass
# constructor will overwrite this with something useful.
self.existing_constraints = {}
logger.debug("%s: logdata %s", self.label, logdata)
def __init__(self, job_uuid):
logger.debug('load stderr events for job %s', job_uuid)
self.job_uuid = job_uuid
def __init__(self, job_uuid):
logger.debug('load stderr events for job %s', job_uuid)
self.job_uuid = job_uuid
def run(self):
logger.debug("%s: parsing logdata %s", self.label, self._logdata)
for line in self._logdata:
m = re.search(r'^\S+ \S+ \d+ (?P<seq>\d+) job_task (?P<task_uuid>\S+)$', line)
if m:
seq = int(m.group('seq'))
uuid = m.group('task_uuid')
self.seq_to_uuid[seq] = uuid
logger.debug('%s: seq %d is task %s', self.label, seq, uuid)
continue
m = re.search(r'^\S+ \S+ \d+ (?P<seq>\d+) (success in|failure \(#., permanent\) after) (?P<elapsed>\d+) seconds', line)
if m:
task_id = self.seq_to_uuid[int(m.group('seq'))]
continue
m = re.search(r'^\S+ \S+ \d+ (?P<seq>\d+) stderr Queued job (?P<uuid>\S+)$', line)
if m:
uuid = m.group('uuid')
logger.debug('%s: follow %s', self.label, uuid)
child_summarizer = JobSummarizer(uuid)
child_summarizer.tasks = self.tasks
child_summarizer.run()
logger.debug('%s: done %s', self.label, uuid)
continue
m = re.search(r'^(?P<timestamp>[^\s.]+)(\.\d+)? (?P<job_uuid>\S+) \d+ (?P<seq>\d+) stderr crunchstat: (?P<category>\S+) (?P<current>.*?)( -- interval (?P<interval>.*))?\n', line)
if m and not self._include_crunchstat_summary:
continue
m = re.search(r'^(?P<timestamp>[^\s.]+)(\.\d+)? (?P<job_uuid>\S+) \d+ (?P<seq>\d+) stderr (?P<log_entry>.*)\n', line)
if not m:
#logger.debug("Could not parse line: %s", line)
continue
if self.label is None:
self.label = m.group('job_uuid')
logger.debug('%s: using job uuid as label', self.label)
task_id = self.seq_to_uuid[int(m.group('seq'))]
task = self.tasks[task_id]
logger.info('[%s] %s', m.group('seq'), m.group('log_entry'))
logger.debug('%s: done parsing', self.label)
def run(self):
logger.debug("%s: parsing logdata %s", self.label, self._logdata)
for line in self._logdata:
m = re.search(
r'^\S+ \S+ \d+ (?P<seq>\d+) job_task (?P<task_uuid>\S+)$',
line)
if m:
seq = int(m.group('seq'))
uuid = m.group('task_uuid')
self.seq_to_uuid[seq] = uuid
logger.debug('%s: seq %d is task %s', self.label, seq, uuid)
continue
m = re.search(
r'^\S+ \S+ \d+ (?P<seq>\d+) (success in|failure \(#., permanent\) after) (?P<elapsed>\d+) seconds',
line)
if m:
task_id = self.seq_to_uuid[int(m.group('seq'))]
continue
m = re.search(
r'^\S+ \S+ \d+ (?P<seq>\d+) stderr Queued job (?P<uuid>\S+)$',
line)
if m:
uuid = m.group('uuid')
logger.debug('%s: follow %s', self.label, uuid)
child_summarizer = JobSummarizer(uuid)
child_summarizer.tasks = self.tasks
child_summarizer.run()
logger.debug('%s: done %s', self.label, uuid)
continue
m = re.search(
r'^(?P<timestamp>[^\s.]+)(\.\d+)? (?P<job_uuid>\S+) \d+ (?P<seq>\d+) stderr crunchstat: (?P<category>\S+) (?P<current>.*?)( -- interval (?P<interval>.*))?\n',
line)
if m and not self._include_crunchstat_summary:
continue
m = re.search(
r'^(?P<timestamp>[^\s.]+)(\.\d+)? (?P<job_uuid>\S+) \d+ (?P<seq>\d+) stderr (?P<log_entry>.*)\n',
line)
if not m:
#logger.debug("Could not parse line: %s", line)
continue
if self.label is None:
self.label = m.group('job_uuid')
logger.debug('%s: using job uuid as label', self.label)
task_id = self.seq_to_uuid[int(m.group('seq'))]
task = self.tasks[task_id]
logger.info('[%s] %s', m.group('seq'), m.group('log_entry'))
logger.debug('%s: done parsing', self.label)