Garbage collector metrics lead to deadlock #363
Description
Story:
We have default gc metrics, for example python_gc_collected_objects_count.
Someone tries to get metrics value(for example http handler). This leads to MutexValue.get() call. Lock is aquired inside get() method.
If garbage collector decides to run at this moment, the callback updating gc metrics is triggered. Inside this callback metrics is need to be updated, but it cant get the lock which is already taken. Deadlock.
Snippet reproducing this story: (if you try to open localhost:8000 it will fail).
import gc
import time
from threading import Lock
from prometheus_client import values
from prometheus_client import start_http_server
class MutexValue(object):
'''A float protected by a mutex.'''
_multiprocess = False
def __init__(self, typ, metric_name, name, labelnames, labelvalues, **kwargs):
self._value = 0.0
self._lock = Lock()
def inc(self, amount):
with self._lock:
self._value += amount
def set(self, value):
with self._lock:
self._value = value
def get(self):
with self._lock:
gc.collect() #<------------------------------ Added only this line
return self._value
values.ValueClass = MutexValue
start_http_server(8000)
while True:
time.sleep(1)Do we really need lock in get() method?