def test_bucketize(self):
""" test bucketize """
junk = defaultdict(list)
data = [1, 1, 1, 1, 1, 1, 1, 1]
junk[datetime.datetime(2019, 5, 11, 0, 0, 0, 0)] = data
junk[datetime.datetime(2019, 5, 11, 1, 0, 0, 0)] = data
junk[datetime.datetime(2019, 5, 11, 2, 0, 0, 0)] = data
junk[datetime.datetime(2019, 5, 11, 3, 0, 0, 0)] = data
self.assertEqual(
len(
util.bucketize(
junk,
start=datetime.datetime(2019, 5, 11),
end=datetime.datetime(2019, 5, 11, 3),
)),
4,
)
self.assertEqual(
len(
util.bucketize(
junk,
start=datetime.datetime(2019, 5, 11),
end=datetime.datetime(2019, 5, 11, 3),
seconds=86400,
)),
1,
)
def print_report(self, interval=3600, by_node=False, top=3):
"""print gc report"""
print("gcinspector version %s" % VERSION)
print("")
if not self.analyzed:
self.analyze()
if not self.pauses:
print("No pauses found")
return
if not by_node:
pauses = self.all_pauses()
self.__print_gc(
sorted(
bucketize(pauses,
start=self.start,
end=self.end,
seconds=interval).items(),
key=lambda t: t[0],
))
plist = []
for time in pauses:
plist.extend(pauses[time])
worst_k = heapq.nlargest(top, plist)
print("Worst pauses in ms:")
print(worst_k)
else:
for node in self.pauses:
print(node)
self.__print_gc(
sorted(
bucketize(
self.pauses[node],
start=self.starts[node],
end=self.ends[node],
seconds=interval,
).items(),
key=lambda t: t[0],
))
plist = []
for time, pauses in self.pauses[node].items():
plist.extend(pauses)
worst_k = heapq.nlargest(top, plist)
print("Worst pauses in ms:")
print(worst_k)
print("")
print("")
print("Collections by type")
print("-" * 20)
for collection, count in self.gc_types.items():
print("* %s: %s" % (collection, count))
print("")
def generate(self, parsed):
"""generates a time series report for a tarball"""
table = []
table.append("")
table.append("filter cache evictions by hour")
table.append("------------------------------")
events_by_datetime = OrderedDefaultDict(list)
start = dates.max_utc_time()
end = dates.min_utc_time()
for node, events in parsed["nodes"].items():
for info in events.get("evictions"):
# put into structure we can use for bucketize
for value in info.values():
if value.time_stamp > end:
end = value.time_stamp
if value.time_stamp < start:
start = value.time_stamp
events_by_datetime[value.time_stamp].append(value)
buckets = sorted(
util.bucketize(events_by_datetime, start, end, 3600).items(),
key=lambda t: t[0],
)
maxval = len(max(buckets, key=lambda t: len(t[1]))[1])
for time, matches in buckets:
pad = ""
for x in range(len(str(maxval)) - len(str(len(matches)))):
pad += " "
table.append("%s %s %s" % (
time.strftime("%Y-%m-%d %H:%M:%S") + pad,
len(matches),
util.textbar(maxval, len(matches)),
))
return "\n".join(table)
def print_report(self, interval=3600):
""" print bucketized result counts """
print("bucketgrep version %s" % VERSION)
print("search: '%s'" % self.supplied_regex)
print()
if not self.analyzed:
self.analyze()
if not self.matches:
print("No matches found")
if self.unknown:
print(self.unknown, "matches without timestamp")
return
buckets = sorted(
bucketize(self.matches,
start=self.start,
end=self.end,
seconds=interval).items(),
key=lambda t: t[0],
)
maxval = len(max(buckets, key=lambda t: len(t[1]))[1])
for time, matches in buckets:
pad = ""
for x in range(len(str(maxval)) - len(str(len(matches)))):
pad += " "
print(
time.strftime("%Y-%m-%d %H:%M:%S") + pad,
len(matches),
textbar(maxval, len(matches)),
)
if self.unknown:
print(self.unknown, "matches without timestamp")
def print_report(self, command_name, interval=3600, top=3):
""" print the report """
if not self.analyzed:
self.analyze()
print("%s version %s" % (command_name, VERSION))
print('')
if not self.queries:
if self.files:
print("no queries found the files provided")
for file_name in self.files:
print("- %s" % file_name)
else:
print("no queries found in diag tarball '%s'" % self.diag_dir)
return
self.__print_query_times(
sorted(bucketize(self.querytimes,
start=self.start,
end=self.end,
seconds=interval).items(),
key=lambda t: t[0]))
print(len(self.queries), "slow queries, %s cross-node" % self.cross)
print()
print("Top %s slow queries:" % top)
print('-' * 30)
for query, time in sorted(self.queries,
key=lambda t: t[1],
reverse=True)[0:top]:
print("%sms: %s" % (time, query))
print('')
def test_bucketize():
""" test bucketize """
junk = defaultdict(list)
data = [1, 1, 1, 1, 1, 1, 1, 1]
junk[datetime.datetime(2019, 5, 11, 0, 0, 0, 0)] = data
junk[datetime.datetime(2019, 5, 11, 1, 0, 0, 0)] = data
junk[datetime.datetime(2019, 5, 11, 2, 0, 0, 0)] = data
junk[datetime.datetime(2019, 5, 11, 3, 0, 0, 0)] = data
assert len(
util.bucketize(junk,
start=datetime.datetime(2019, 5, 11),
end=datetime.datetime(2019, 5, 11, 3))) == 4
assert len(
util.bucketize(junk,
start=datetime.datetime(2019, 5, 11),
end=datetime.datetime(2019, 5, 11, 3),
seconds=86400)) == 1
def print_report(self, command_name, interval=3600, top=3):
"""print the report"""
if not self.analyzed:
self.analyze()
print("%s version: %s" % (command_name, VERSION))
print("")
print(
"this is not a very accurate report, use it to discover basics, but I suggest analyzing the logs by hand for any outliers"
)
print("")
if not self.queries:
if self.files:
print("no queries found the files provided")
for file_name in self.files:
print("- %s" % file_name)
else:
print("no queries found in diag tarball '%s'" % self.diag_dir)
return
self.__print_query_times(
sorted(
bucketize(self.querytimes,
start=self.start,
end=self.end,
seconds=interval).items(),
key=lambda t: t[0],
))
print("slow query breakdown")
print("--------------------")
print(
len(self.queries),
"total, %s cross-node, %s timeouts" % (self.cross, self.timedout),
)
print()
print("Top %s slow queries:" % top)
print("-" * 30)
for query, time in sorted(self.queries,
key=lambda t: (t[1], t[0]),
reverse=True)[0:top]:
print("%sms: %s" % (time, query))
print("")
def print_report(self, interval=3600):
"""print bucketized result counts"""
print()
if not self.analyzed:
self.analyze()
if not self.matches:
print("No matches found")
if self.unknown:
print(self.unknown, "matches without timestamp")
return
if self.report == "summary":
print()
print("cluster wide")
print("------------")
buckets = sorted(
bucketize(self.matches,
start=self.start,
end=self.end,
seconds=interval).items(),
key=lambda t: t[0],
)
maxval = len(max(buckets, key=lambda t: len(t[1]))[1])
for time, matches in buckets:
pad = ""
for x in range(len(str(maxval)) - len(str(len(matches)))):
pad += " "
print(
time.strftime("%Y-%m-%d %H:%M:%S") + pad,
len(matches),
textbar(maxval, len(matches)),
)
else:
print()
print()
print("per node numbers")
print("----------------")
for node in sorted(self.node_matches.keys()):
print()
print("node: %s" % node)
print("--------")
if not len(self.node_matches[node]):
print("No matches for %s found" % node)
continue
buckets = sorted(
bucketize(
self.node_matches[node],
start=self.start,
end=self.end,
seconds=interval,
).items(),
key=lambda t: t[0],
)
maxval = len(max(buckets, key=lambda t: len(t[1]))[1])
for time, matches in buckets:
pad = ""
for x in range(len(str(maxval)) - len(str(len(matches)))):
pad += " "
print(
time.strftime("%Y-%m-%d %H:%M:%S") + pad,
len(matches),
textbar(maxval, len(matches)),
)
if self.unknown:
print(self.unknown, "matches without timestamp")