def matcher(stat, data):
hostname = stat["hostname"]
job = stat["job_name"]
phase = stat["phase_name"]
epoch = stat["epoch"]
if "stage_name" in stat:
stage_name = stat["stage_name"]
colon_index = stage_name.find(':')
if colon_index != -1:
stage_name = stage_name[:colon_index]
worker_id = stat["id"]
else:
stage_name = stat["logger_name"]
worker_id = None
val = stat[stat_value_key]
# Add value to global job statistics
global_data_key_list = [
"stats", (job, phase, epoch), (stage_name, ), data_key
]
global_stat_container = utils.populate_nested_dictionary(
data,
key_list=global_data_key_list,
types=[dict, dict, dict, StatContainer])
global_stat_container.append(val)
# Add value to per-host statistics
per_host_data_key_list = [
"stats", (job, phase, epoch), (hostname, stage_name), data_key
]
per_host_stat_container = utils.populate_nested_dictionary(
data,
key_list=per_host_data_key_list,
types=[dict, dict, dict, StatContainer])
per_host_stat_container.append(val)
# If this is a statistic pertaining to a worker, add value to per-worker
# statistics
if worker_id is not None:
per_worker_data_key_list = [
"stats", (job, phase, epoch),
(hostname, stage_name, worker_id), data_key
]
per_worker_stat_container = utils.populate_nested_dictionary(
data,
key_list=per_worker_data_key_list,
types=[dict, dict, dict, StatContainer])
per_worker_stat_container.append(val)
def matcher(stat, data):
hostname = stat["hostname"]
job = stat["job_name"]
phase = stat["phase_name"]
epoch = stat["epoch"]
if "stage_name" in stat:
stage_name = stat["stage_name"]
colon_index = stage_name.find(':')
if colon_index != -1:
stage_name = stage_name[:colon_index]
worker_id = stat["id"]
else:
stage_name = stat["logger_name"]
worker_id = None
val = stat[stat_value_key]
# Add value to global job statistics
global_data_key_list = [
"stats", (job, phase, epoch), (stage_name,), data_key]
global_stat_container = utils.populate_nested_dictionary(
data, key_list=global_data_key_list,
types=[dict, dict, dict, StatContainer])
global_stat_container.append(val)
# Add value to per-host statistics
per_host_data_key_list = [
"stats", (job, phase, epoch), (hostname, stage_name), data_key]
per_host_stat_container = utils.populate_nested_dictionary(
data, key_list=per_host_data_key_list,
types=[dict, dict, dict, StatContainer])
per_host_stat_container.append(val)
# If this is a statistic pertaining to a worker, add value to per-worker
# statistics
if worker_id is not None:
per_worker_data_key_list = [
"stats", (job, phase, epoch), (hostname, stage_name, worker_id),
data_key]
per_worker_stat_container = utils.populate_nested_dictionary(
data, key_list=per_worker_data_key_list,
types=[dict, dict, dict, StatContainer])
per_worker_stat_container.append(val)
def gather_timestamped_points_matcher(query_number, match, data, data_key):
timestamp = float(match["timestamp"])
phase = match["phase_name"]
stat_name = match["collection_stat_name"]
stat_val = match["value"]
# Update minimum and maximum timestamps
for key, comparison_function in [("min_timestamp", min),
("max_timestamp", max)]:
if key not in data:
data[key] = {}
if phase not in data[key]:
data[key][phase] = timestamp
else:
data[key][phase] = comparison_function(
data[key][phase], timestamp)
# Subdivide points by query, then further by stat name, and worker
# identification so that we can make fine-grained plots if needed
data_subdict = metaprogram_utils.populate_nested_dictionary(
data, ["plot_points", query_number, stat_name, data_key])
for key in ["x_values", "y_values"]:
if key not in data_subdict:
data_subdict[key] = []
data_subdict["x_values"].append(timestamp)
data_subdict["y_values"].append(stat_val)
def set_stage_value_matcher(stat, data):
"""
Operating under the assumption that an entire stage's workers share one
value for a statistic, this matcher will log the statistic for that stage,
overwriting previous values if they exist
"""
job = stat["job_name"]
phase = stat["phase_name"]
epoch = stat["epoch"]
stage_name = stat["stage_name"]
stat_name = stat["stat_name"]
colon_index = stage_name.find(':')
if colon_index != -1:
stage_name = stage_name[:colon_index]
if "str_value" in stat:
stat_val = stat["str_value"]
else:
stat_val = stat["uint_value"]
data_subdict = utils.populate_nested_dictionary(
data, ["stage_info", (job, phase, epoch), stage_name])
data_subdict[stat_name] = stat_val
def gather_timestamped_points_matcher(query_number, match, data, data_key):
timestamp = float(match["timestamp"])
phase = match["phase_name"]
stat_name = match["collection_stat_name"]
stat_val = match["value"]
# Update minimum and maximum timestamps
for key, comparison_function in [("min_timestamp", min), ("max_timestamp", max)]:
if key not in data:
data[key] = {}
if phase not in data[key]:
data[key][phase] = timestamp
else:
data[key][phase] = comparison_function(data[key][phase], timestamp)
# Subdivide points by query, then further by stat name, and worker
# identification so that we can make fine-grained plots if needed
data_subdict = metaprogram_utils.populate_nested_dictionary(
data, ["plot_points", query_number, stat_name, data_key]
)
for key in ["x_values", "y_values"]:
if key not in data_subdict:
data_subdict[key] = []
data_subdict["x_values"].append(timestamp)
data_subdict["y_values"].append(stat_val)
def set_stage_value_matcher(stat, data):
"""
Operating under the assumption that an entire stage's workers share one
value for a statistic, this matcher will log the statistic for that stage,
overwriting previous values if they exist
"""
job = stat["job_name"]
phase = stat["phase_name"]
epoch = stat["epoch"]
stage_name = stat["stage_name"]
stat_name = stat["stat_name"]
colon_index = stage_name.find(':')
if colon_index != -1:
stage_name = stage_name[:colon_index]
if "str_value" in stat:
stat_val = stat["str_value"]
else:
stat_val = stat["uint_value"]
data_subdict = utils.populate_nested_dictionary(
data, ["stage_info", (job, phase, epoch), stage_name])
data_subdict[stat_name] = stat_val
def handleDiskCountMatch(match, data):
job = match["job_name"]
stat_name = match["stat_name"]
hostname = match["hostname"]
disks_subdict = utils.populate_nested_dictionary(
data, [(job, None, None), "disks", hostname])
disks_subdict[stat_name] = match["uint_value"]
def handleDiskCountMatch(match, data):
job = match["job_name"]
stat_name = match["stat_name"]
hostname = match["hostname"]
disks_subdict = utils.populate_nested_dictionary(
data, [(job, None, None), "disks", hostname])
disks_subdict[stat_name] = match["uint_value"]
def handleTimestampQueryMatch(match, data):
job = match["job_name"]
phase = match["phase_name"]
epoch = match["epoch"]
start_time = match["start_time"]
stop_time = match["stop_time"]
start_time_subdict = utils.populate_nested_dictionary(
data,
key_list=[(job, phase, epoch), "stats", "timestamps", "start"],
types=[dict, dict, dict, StatContainer])
start_time_subdict.append(start_time)
stop_time_subdict = utils.populate_nested_dictionary(
data,
key_list=[(job, phase, epoch), "stats", "timestamps", "stop"],
types=[dict, dict, dict, StatContainer])
stop_time_subdict.append(stop_time)
def handleTimestampQueryMatch(match, data):
job = match["job_name"]
phase = match["phase_name"]
epoch = match["epoch"]
start_time = match["start_time"]
stop_time = match["stop_time"]
start_time_subdict = utils.populate_nested_dictionary(
data,
key_list=[(job, phase, epoch), "stats", "timestamps", "start"],
types=[dict, dict, dict, StatContainer])
start_time_subdict.append(start_time)
stop_time_subdict = utils.populate_nested_dictionary(
data,
key_list=[(job, phase, epoch), "stats", "timestamps", "stop"],
types=[dict, dict, dict, StatContainer])
stop_time_subdict.append(stop_time)
def handleReaderInputMatch(match, data):
job = match["job_name"]
phase_name = match["phase_name"]
epoch = match["epoch"]
value = match["uint_value"]
hostname = match["hostname"]
hosts_subdict = utils.populate_nested_dictionary(
data,
key_list=[(job, phase_name, epoch), "hosts"],
types=[dict, set])
hosts_subdict.add(hostname)
data_subdict = utils.populate_nested_dictionary(
data, [(job, phase_name, epoch), "stats"])
if "input_size" not in data_subdict:
data_subdict["input_size"] = 0
data_subdict["input_size"] += value
def handleWriterOutputMatch(match, data):
job = match["job_name"]
phase_name = match["phase_name"]
epoch = match["epoch"]
value = match["uint_value"]
data_subdict = utils.populate_nested_dictionary(
data, [(job, phase_name, epoch), "stats"])
if "output_size" not in data_subdict:
data_subdict["output_size"] = 0
data_subdict["output_size"] += value
def handleReaderInputMatch(match, data):
job = match["job_name"]
phase_name = match["phase_name"]
epoch = match["epoch"]
value = match["uint_value"]
hostname = match["hostname"]
hosts_subdict = utils.populate_nested_dictionary(data,
key_list=[
(job, phase_name,
epoch), "hosts"
],
types=[dict, set])
hosts_subdict.add(hostname)
data_subdict = utils.populate_nested_dictionary(
data, [(job, phase_name, epoch), "stats"])
if "input_size" not in data_subdict:
data_subdict["input_size"] = 0
data_subdict["input_size"] += value
def handleWriterOutputMatch(match, data):
job = match["job_name"]
phase_name = match["phase_name"]
epoch = match["epoch"]
value = match["uint_value"]
data_subdict = utils.populate_nested_dictionary(
data, [(job, phase_name, epoch), "stats"])
if "output_size" not in data_subdict:
data_subdict["output_size"] = 0
data_subdict["output_size"] += value
def gather_histogram_points_matcher(query_number, match, data, data_key):
phase = match["phase_name"]
stat_name = match["stat_name"]
bin_value = match["bin"]
count = match["count"]
data_subdict = metaprogram_utils.populate_nested_dictionary(
data, ["plot_points", query_number, stat_name, data_key])
for key in ["bin", "count"]:
if key not in data_subdict:
data_subdict[key] = []
data_subdict["bin"].append(bin_value)
data_subdict["count"].append(count)
def gather_histogram_points_matcher(query_number, match, data, data_key):
phase = match["phase_name"]
stat_name = match["stat_name"]
bin_value = match["bin"]
count = match["count"]
data_subdict = metaprogram_utils.populate_nested_dictionary(
data, ["plot_points", query_number, stat_name, data_key]
)
for key in ["bin", "count"]:
if key not in data_subdict:
data_subdict[key] = []
data_subdict["bin"].append(bin_value)
data_subdict["count"].append(count)
