diff options
author | Amber Brown <hawkowl@atleastfornow.net> | 2018-09-03 21:08:35 +1000 |
---|---|---|
committer | GitHub <noreply@github.com> | 2018-09-03 21:08:35 +1000 |
commit | 4fc4b881c58fd638db5f4dac0863721111b67af0 (patch) | |
tree | cc1604f5e3b4e0a263e0e11a55b62ef4006a64a1 /synapse/util/logcontext.py | |
parent | The project `matrix-synapse-auto-deploy` does not seem to be maintained anymore. (diff) | |
parent | Merge pull request #3777 from matrix-org/neilj/fix_register_user_registration (diff) | |
download | synapse-4fc4b881c58fd638db5f4dac0863721111b67af0.tar.xz |
Merge branch 'develop' into develop
Diffstat (limited to 'synapse/util/logcontext.py')
-rw-r--r-- | synapse/util/logcontext.py | 201 |
1 files changed, 166 insertions, 35 deletions
diff --git a/synapse/util/logcontext.py b/synapse/util/logcontext.py index a58c723403..a0c2d37610 100644 --- a/synapse/util/logcontext.py +++ b/synapse/util/logcontext.py @@ -22,10 +22,10 @@ them. See doc/log_contexts.rst for details on how this works. """ -from twisted.internet import defer - -import threading import logging +import threading + +from twisted.internet import defer logger = logging.getLogger(__name__) @@ -49,17 +49,118 @@ except Exception: return None +class ContextResourceUsage(object): + """Object for tracking the resources used by a log context + + Attributes: + ru_utime (float): user CPU time (in seconds) + ru_stime (float): system CPU time (in seconds) + db_txn_count (int): number of database transactions done + db_sched_duration_sec (float): amount of time spent waiting for a + database connection + db_txn_duration_sec (float): amount of time spent doing database + transactions (excluding scheduling time) + evt_db_fetch_count (int): number of events requested from the database + """ + + __slots__ = [ + "ru_stime", "ru_utime", + "db_txn_count", "db_txn_duration_sec", "db_sched_duration_sec", + "evt_db_fetch_count", + ] + + def __init__(self, copy_from=None): + """Create a new ContextResourceUsage + + Args: + copy_from (ContextResourceUsage|None): if not None, an object to + copy stats from + """ + if copy_from is None: + self.reset() + else: + self.ru_utime = copy_from.ru_utime + self.ru_stime = copy_from.ru_stime + self.db_txn_count = copy_from.db_txn_count + + self.db_txn_duration_sec = copy_from.db_txn_duration_sec + self.db_sched_duration_sec = copy_from.db_sched_duration_sec + self.evt_db_fetch_count = copy_from.evt_db_fetch_count + + def copy(self): + return ContextResourceUsage(copy_from=self) + + def reset(self): + self.ru_stime = 0. + self.ru_utime = 0. + self.db_txn_count = 0 + + self.db_txn_duration_sec = 0 + self.db_sched_duration_sec = 0 + self.evt_db_fetch_count = 0 + + def __repr__(self): + return ("<ContextResourceUsage ru_stime='%r', ru_utime='%r', " + "db_txn_count='%r', db_txn_duration_sec='%r', " + "db_sched_duration_sec='%r', evt_db_fetch_count='%r'>") % ( + self.ru_stime, + self.ru_utime, + self.db_txn_count, + self.db_txn_duration_sec, + self.db_sched_duration_sec, + self.evt_db_fetch_count,) + + def __iadd__(self, other): + """Add another ContextResourceUsage's stats to this one's. + + Args: + other (ContextResourceUsage): the other resource usage object + """ + self.ru_utime += other.ru_utime + self.ru_stime += other.ru_stime + self.db_txn_count += other.db_txn_count + self.db_txn_duration_sec += other.db_txn_duration_sec + self.db_sched_duration_sec += other.db_sched_duration_sec + self.evt_db_fetch_count += other.evt_db_fetch_count + return self + + def __isub__(self, other): + self.ru_utime -= other.ru_utime + self.ru_stime -= other.ru_stime + self.db_txn_count -= other.db_txn_count + self.db_txn_duration_sec -= other.db_txn_duration_sec + self.db_sched_duration_sec -= other.db_sched_duration_sec + self.evt_db_fetch_count -= other.evt_db_fetch_count + return self + + def __add__(self, other): + res = ContextResourceUsage(copy_from=self) + res += other + return res + + def __sub__(self, other): + res = ContextResourceUsage(copy_from=self) + res -= other + return res + + class LoggingContext(object): """Additional context for log formatting. Contexts are scoped within a "with" block. + If a parent is given when creating a new context, then: + - logging fields are copied from the parent to the new context on entry + - when the new context exits, the cpu usage stats are copied from the + child to the parent + Args: name (str): Name for the context for debugging. + parent_context (LoggingContext|None): The parent of the new context """ __slots__ = [ - "previous_context", "name", "ru_stime", "ru_utime", - "db_txn_count", "db_txn_duration_sec", "db_sched_duration_sec", + "previous_context", "name", "parent_context", + "_resource_usage", "usage_start", "main_thread", "alive", "request", "tag", @@ -90,24 +191,21 @@ class LoggingContext(object): def add_database_scheduled(self, sched_sec): pass + def record_event_fetch(self, event_count): + pass + def __nonzero__(self): return False __bool__ = __nonzero__ # python3 sentinel = Sentinel() - def __init__(self, name=None): + def __init__(self, name=None, parent_context=None): self.previous_context = LoggingContext.current_context() self.name = name - self.ru_stime = 0. - self.ru_utime = 0. - self.db_txn_count = 0 - - # sec spent waiting for db txns, excluding scheduling time - self.db_txn_duration_sec = 0 - # sec spent waiting for db txns to be scheduled - self.db_sched_duration_sec = 0 + # track the resources used by this context so far + self._resource_usage = ContextResourceUsage() # If alive has the thread resource usage when the logcontext last # became active. @@ -118,6 +216,8 @@ class LoggingContext(object): self.tag = "" self.alive = True + self.parent_context = parent_context + def __str__(self): return "%s@%x" % (self.name, id(self)) @@ -155,6 +255,10 @@ class LoggingContext(object): self.previous_context, old_context ) self.alive = True + + if self.parent_context is not None: + self.parent_context.copy_to(self) + return self def __exit__(self, type, value, traceback): @@ -176,6 +280,13 @@ class LoggingContext(object): self.previous_context = None self.alive = False + # if we have a parent, pass our CPU usage stats on + if self.parent_context is not None: + self.parent_context._resource_usage += self._resource_usage + + # reset them in case we get entered again + self._resource_usage.reset() + def copy_to(self, record): """Copy logging fields from this context to a log record or another LoggingContext @@ -200,39 +311,43 @@ class LoggingContext(object): logger.warning("Stopped logcontext %s on different thread", self) return - # When we stop, let's record the resource used since we started - if self.usage_start: - usage_end = get_thread_resource_usage() + # When we stop, let's record the cpu used since we started + if not self.usage_start: + logger.warning( + "Called stop on logcontext %s without calling start", self, + ) + return + + usage_end = get_thread_resource_usage() - self.ru_utime += usage_end.ru_utime - self.usage_start.ru_utime - self.ru_stime += usage_end.ru_stime - self.usage_start.ru_stime + self._resource_usage.ru_utime += usage_end.ru_utime - self.usage_start.ru_utime + self._resource_usage.ru_stime += usage_end.ru_stime - self.usage_start.ru_stime - self.usage_start = None - else: - logger.warning("Called stop on logcontext %s without calling start", self) + self.usage_start = None def get_resource_usage(self): - """Get CPU time used by this logcontext so far. + """Get resources used by this logcontext so far. Returns: - tuple[float, float]: The user and system CPU usage in seconds + ContextResourceUsage: a *copy* of the object tracking resource + usage so far """ - ru_utime = self.ru_utime - ru_stime = self.ru_stime + # we always return a copy, for consistency + res = self._resource_usage.copy() # If we are on the correct thread and we're currently running then we # can include resource usage so far. is_main_thread = threading.current_thread() is self.main_thread if self.alive and self.usage_start and is_main_thread: current = get_thread_resource_usage() - ru_utime += current.ru_utime - self.usage_start.ru_utime - ru_stime += current.ru_stime - self.usage_start.ru_stime + res.ru_utime += current.ru_utime - self.usage_start.ru_utime + res.ru_stime += current.ru_stime - self.usage_start.ru_stime - return ru_utime, ru_stime + return res def add_database_transaction(self, duration_sec): - self.db_txn_count += 1 - self.db_txn_duration_sec += duration_sec + self._resource_usage.db_txn_count += 1 + self._resource_usage.db_txn_duration_sec += duration_sec def add_database_scheduled(self, sched_sec): """Record a use of the database pool @@ -241,7 +356,15 @@ class LoggingContext(object): sched_sec (float): number of seconds it took us to get a connection """ - self.db_sched_duration_sec += sched_sec + self._resource_usage.db_sched_duration_sec += sched_sec + + def record_event_fetch(self, event_count): + """Record a number of events being fetched from the db + + Args: + event_count (int): number of events being fetched + """ + self._resource_usage.evt_db_fetch_count += event_count class LoggingContextFilter(logging.Filter): @@ -262,7 +385,13 @@ class LoggingContextFilter(logging.Filter): context = LoggingContext.current_context() for key, value in self.defaults.items(): setattr(record, key, value) - context.copy_to(record) + + # context should never be None, but if it somehow ends up being, then + # we end up in a death spiral of infinite loops, so let's check, for + # robustness' sake. + if context is not None: + context.copy_to(record) + return True @@ -273,7 +402,9 @@ class PreserveLoggingContext(object): __slots__ = ["current_context", "new_context", "has_parent"] - def __init__(self, new_context=LoggingContext.sentinel): + def __init__(self, new_context=None): + if new_context is None: + new_context = LoggingContext.sentinel self.new_context = new_context def __enter__(self): @@ -403,7 +534,7 @@ _to_ignore = [ "synapse.util.logcontext", "synapse.http.server", "synapse.storage._base", - "synapse.util.async", + "synapse.util.async_helpers", ] |