def logs(obj, input_path, stdout=None, stderr=None, both=None):
types = set()
if stdout:
types.add('stdout')
both = False
if stderr:
types.add('stderr')
both = False
if both:
types.update(('stdout', 'stderr'))
# Pathspec can either be run_id/step_name or run_id/step_name/task_id.
parts = input_path.split('/')
if len(parts) == 2:
run_id, step_name = parts
task_id = None
elif len(parts) == 3:
run_id, step_name, task_id = parts
else:
raise CommandException("input_path should either be run_id/step_name"
"or run_id/step_name/task_id")
if obj.datastore.datastore_root is None:
obj.datastore.datastore_root = obj.datastore.get_datastore_root_from_config(
obj.echo, create_on_absent=False)
if obj.datastore.datastore_root is None:
raise CommandException(
"Could not find the location of the datastore -- did you correctly set the "
"METAFLOW_DATASTORE_SYSROOT_%s environment variable" %
(obj.datastore.TYPE).upper())
from metaflow.datastore.datastore_set import MetaflowDatastoreSet
datastore_set = MetaflowDatastoreSet(obj.datastore,
obj.flow.name,
run_id,
steps=[step_name],
metadata=obj.metadata,
monitor=obj.monitor,
event_logger=obj.event_logger)
if task_id:
ds_list = [datastore_set.get_with_pathspec(input_path)]
else:
ds_list = list(datastore_set) # get all tasks
for ds in ds_list:
echo('Dumping logs of run_id=*{run_id}* '
'step=*{step}* task_id=*{task_id}*'.format(run_id=ds.run_id,
step=ds.step_name,
task_id=ds.task_id),
fg='magenta')
for typ in ('stdout', 'stderr'):
if typ in types:
echo(typ, bold=True)
click.secho(ds.load_log(typ).decode('UTF-8', errors='replace'),
nl=False)
def dump(obj,
input_path,
private=None,
max_value_size=None,
include=None,
file=None):
output = {}
kwargs = {
'show_private': private,
'max_value_size': max_value_size,
'include': {t
for t in include.split(',') if t}
}
if obj.datastore.datastore_root is None:
obj.datastore.datastore_root = obj.datastore.get_datastore_root_from_config(
obj.echo, create_on_absent=False)
if obj.datastore.datastore_root is None:
raise CommandException(
"Could not find the location of the datastore -- did you correctly set the "
"METAFLOW_DATASTORE_SYSROOT_%s environment variable" %
(obj.datastore.TYPE).upper())
# Pathspec can either be run_id/step_name or run_id/step_name/task_id.
parts = input_path.split('/')
if len(parts) == 2:
run_id, step_name = parts
task_id = None
elif len(parts) == 3:
run_id, step_name, task_id = parts
else:
raise CommandException("input_path should either be run_id/step_name"
"or run_id/step_name/task_id")
from metaflow.datastore.datastore_set import MetaflowDatastoreSet
datastore_set = MetaflowDatastoreSet(
obj.datastore,
obj.flow.name,
run_id,
steps=[step_name],
metadata=obj.metadata,
monitor=obj.monitor,
event_logger=obj.event_logger,
prefetch_data_artifacts=kwargs.get('include'))
if task_id:
ds_list = [datastore_set.get_with_pathspec(input_path)]
else:
ds_list = list(datastore_set) # get all tasks
for ds in ds_list:
echo('Dumping output of run_id=*{run_id}* '
'step=*{step}* task_id=*{task_id}*'.format(run_id=ds.run_id,
step=ds.step_name,
task_id=ds.task_id),
fg='magenta')
if file is None:
echo_always(ds.format(**kwargs),
highlight='green',
highlight_bold=False,
err=False)
else:
output[ds.pathspec] = ds.to_dict(**kwargs)
if file is not None:
with open(file, 'wb') as f:
pickle.dump(output, f, protocol=pickle.HIGHEST_PROTOCOL)
echo('Artifacts written to *%s*' % file)
def logs(obj,
input_path,
stdout=None,
stderr=None,
both=None,
timestamps=False):
types = set()
if stdout:
types.add('stdout')
both = False
if stderr:
types.add('stderr')
both = False
if both:
types.update(('stdout', 'stderr'))
streams = list(sorted(types, reverse=True))
# Pathspec can either be run_id/step_name or run_id/step_name/task_id.
parts = input_path.split('/')
if len(parts) == 2:
run_id, step_name = parts
task_id = None
elif len(parts) == 3:
run_id, step_name, task_id = parts
else:
raise CommandException("input_path should either be run_id/step_name "
"or run_id/step_name/task_id")
if obj.datastore.datastore_root is None:
obj.datastore.datastore_root = obj.datastore.get_datastore_root_from_config(
obj.echo, create_on_absent=False)
if obj.datastore.datastore_root is None:
raise CommandException(
"Could not find the location of the datastore -- did you correctly set the "
"METAFLOW_DATASTORE_SYSROOT_%s environment variable" %
(obj.datastore.TYPE).upper())
if task_id:
ds_list = [
obj.datastore(obj.flow.name,
run_id=run_id,
step_name=step_name,
task_id=task_id,
mode='r',
allow_unsuccessful=True)
]
else:
from metaflow.datastore.datastore_set import MetaflowDatastoreSet
datastore_set = MetaflowDatastoreSet(obj.datastore,
obj.flow.name,
run_id,
steps=[step_name],
metadata=obj.metadata,
monitor=obj.monitor,
event_logger=obj.event_logger)
# get all successful tasks
ds_list = list(datastore_set)
if ds_list:
def echo_unicode(line, **kwargs):
click.secho(line.decode('UTF-8', errors='replace'), **kwargs)
# old style logs are non mflog-style logs
maybe_old_style = True
for ds in ds_list:
echo('Dumping logs of run_id=*{run_id}* '
'step=*{step}* task_id=*{task_id}*'.format(
run_id=ds.run_id, step=ds.step_name, task_id=ds.task_id),
fg='magenta')
for stream in streams:
echo(stream, bold=True)
logs = ds.load_logs(LOG_SOURCES, stream)
if any(data for _, data in logs):
# attempt to read new, mflog-style logs
for line in mflog.merge_logs([blob for _, blob in logs]):
if timestamps:
ts = mflog.utc_to_local(line.utc_tstamp)
tstamp = ts.strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]
click.secho(tstamp + ' ',
fg=LOGGER_TIMESTAMP,
nl=False)
echo_unicode(line.msg)
maybe_old_style = False
elif maybe_old_style:
# if they are not available, we may be looking at
# a legacy run (unless we have seen new-style data already
# for another stream). This return an empty string if
# nothing is found
log = ds.load_log_legacy(stream)
if log and timestamps:
raise CommandException(
"We can't show --timestamps for "
"old runs. Sorry!")
echo_unicode(log, nl=False)
elif len(parts) == 2:
# TODO if datastore provided a way to find unsuccessful task IDs, we
# could make handle this case automatically
raise CommandException("Successful tasks were not found at the given "
"path. You can see logs for unsuccessful tasks "
"by giving an exact task ID using the "
"run_id/step_name/task_id format.")
