From 20c6d85c6e8f721b8688b7f1361c7a7bab2449fd Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 13 Jan 2022 14:35:52 +0000 Subject: [PATCH] Simplify GC prometheus metrics (#11723) Rather than hooking into the reactor loop, just add a timed task that runs every 100 ms to do the garbage collection. Part 1 of a quest to simplify the reactor monkey-patching. --- changelog.d/11723.misc | 1 + synapse/app/_base.py | 3 +- synapse/metrics/__init__.py | 162 +--------------------------- synapse/metrics/_gc.py | 203 ++++++++++++++++++++++++++++++++++++ 4 files changed, 209 insertions(+), 160 deletions(-) create mode 100644 changelog.d/11723.misc create mode 100644 synapse/metrics/_gc.py diff --git a/changelog.d/11723.misc b/changelog.d/11723.misc new file mode 100644 index 000000000..f99e02070 --- /dev/null +++ b/changelog.d/11723.misc @@ -0,0 +1 @@ +Simplify calculation of prometheus metrics for garbage collection. diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 5fc59c1be..579adbbca 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -60,7 +60,7 @@ from synapse.events.spamcheck import load_legacy_spam_checkers from synapse.events.third_party_rules import load_legacy_third_party_event_rules from synapse.handlers.auth import load_legacy_password_auth_providers from synapse.logging.context import PreserveLoggingContext -from synapse.metrics import register_threadpool +from synapse.metrics import install_gc_manager, register_threadpool from synapse.metrics.background_process_metrics import wrap_as_background_process from synapse.metrics.jemalloc import setup_jemalloc_stats from synapse.types import ISynapseReactor @@ -159,6 +159,7 @@ def start_reactor( change_resource_limit(soft_file_limit) if gc_thresholds: gc.set_threshold(*gc_thresholds) + install_gc_manager() run_command() # make sure that we run the reactor with the sentinel log context, diff --git a/synapse/metrics/__init__.py b/synapse/metrics/__init__.py index 269c2b989..ba7ca0f2d 100644 --- a/synapse/metrics/__init__.py +++ b/synapse/metrics/__init__.py @@ -13,7 +13,6 @@ # limitations under the License. import functools -import gc import itertools import logging import os @@ -41,7 +40,6 @@ import attr from prometheus_client import CollectorRegistry, Counter, Gauge, Histogram, Metric from prometheus_client.core import ( REGISTRY, - CounterMetricFamily, GaugeHistogramMetricFamily, GaugeMetricFamily, ) @@ -56,13 +54,13 @@ from synapse.metrics._exposition import ( generate_latest, start_http_server, ) +from synapse.metrics._gc import MIN_TIME_BETWEEN_GCS, install_gc_manager from synapse.util.versionstring import get_version_string logger = logging.getLogger(__name__) METRICS_PREFIX = "/_synapse/metrics" -running_on_pypy = platform.python_implementation() == "PyPy" all_gauges: "Dict[str, Union[LaterGauge, InFlightGauge]]" = {} HAVE_PROC_SELF_STAT = os.path.exists("/proc/self/stat") @@ -369,121 +367,6 @@ class CPUMetrics: REGISTRY.register(CPUMetrics()) -# -# Python GC metrics -# - -gc_unreachable = Gauge("python_gc_unreachable_total", "Unreachable GC objects", ["gen"]) -gc_time = Histogram( - "python_gc_time", - "Time taken to GC (sec)", - ["gen"], - buckets=[ - 0.0025, - 0.005, - 0.01, - 0.025, - 0.05, - 0.10, - 0.25, - 0.50, - 1.00, - 2.50, - 5.00, - 7.50, - 15.00, - 30.00, - 45.00, - 60.00, - ], -) - - -class GCCounts: - def collect(self) -> Iterable[Metric]: - cm = GaugeMetricFamily("python_gc_counts", "GC object counts", labels=["gen"]) - for n, m in enumerate(gc.get_count()): - cm.add_metric([str(n)], m) - - yield cm - - -if not running_on_pypy: - REGISTRY.register(GCCounts()) - - -# -# PyPy GC / memory metrics -# - - -class PyPyGCStats: - def collect(self) -> Iterable[Metric]: - - # @stats is a pretty-printer object with __str__() returning a nice table, - # plus some fields that contain data from that table. - # unfortunately, fields are pretty-printed themselves (i. e. '4.5MB'). - stats = gc.get_stats(memory_pressure=False) # type: ignore - # @s contains same fields as @stats, but as actual integers. - s = stats._s # type: ignore - - # also note that field naming is completely braindead - # and only vaguely correlates with the pretty-printed table. - # >>>> gc.get_stats(False) - # Total memory consumed: - # GC used: 8.7MB (peak: 39.0MB) # s.total_gc_memory, s.peak_memory - # in arenas: 3.0MB # s.total_arena_memory - # rawmalloced: 1.7MB # s.total_rawmalloced_memory - # nursery: 4.0MB # s.nursery_size - # raw assembler used: 31.0kB # s.jit_backend_used - # ----------------------------- - # Total: 8.8MB # stats.memory_used_sum - # - # Total memory allocated: - # GC allocated: 38.7MB (peak: 41.1MB) # s.total_allocated_memory, s.peak_allocated_memory - # in arenas: 30.9MB # s.peak_arena_memory - # rawmalloced: 4.1MB # s.peak_rawmalloced_memory - # nursery: 4.0MB # s.nursery_size - # raw assembler allocated: 1.0MB # s.jit_backend_allocated - # ----------------------------- - # Total: 39.7MB # stats.memory_allocated_sum - # - # Total time spent in GC: 0.073 # s.total_gc_time - - pypy_gc_time = CounterMetricFamily( - "pypy_gc_time_seconds_total", - "Total time spent in PyPy GC", - labels=[], - ) - pypy_gc_time.add_metric([], s.total_gc_time / 1000) - yield pypy_gc_time - - pypy_mem = GaugeMetricFamily( - "pypy_memory_bytes", - "Memory tracked by PyPy allocator", - labels=["state", "class", "kind"], - ) - # memory used by JIT assembler - pypy_mem.add_metric(["used", "", "jit"], s.jit_backend_used) - pypy_mem.add_metric(["allocated", "", "jit"], s.jit_backend_allocated) - # memory used by GCed objects - pypy_mem.add_metric(["used", "", "arenas"], s.total_arena_memory) - pypy_mem.add_metric(["allocated", "", "arenas"], s.peak_arena_memory) - pypy_mem.add_metric(["used", "", "rawmalloced"], s.total_rawmalloced_memory) - pypy_mem.add_metric(["allocated", "", "rawmalloced"], s.peak_rawmalloced_memory) - pypy_mem.add_metric(["used", "", "nursery"], s.nursery_size) - pypy_mem.add_metric(["allocated", "", "nursery"], s.nursery_size) - # totals - pypy_mem.add_metric(["used", "totals", "gc"], s.total_gc_memory) - pypy_mem.add_metric(["allocated", "totals", "gc"], s.total_allocated_memory) - pypy_mem.add_metric(["used", "totals", "gc_peak"], s.peak_memory) - pypy_mem.add_metric(["allocated", "totals", "gc_peak"], s.peak_allocated_memory) - yield pypy_mem - - -if running_on_pypy: - REGISTRY.register(PyPyGCStats()) - # # Twisted reactor metrics @@ -612,14 +495,6 @@ class ReactorLastSeenMetric: REGISTRY.register(ReactorLastSeenMetric()) -# The minimum time in seconds between GCs for each generation, regardless of the current GC -# thresholds and counts. -MIN_TIME_BETWEEN_GCS = (1.0, 10.0, 30.0) - -# The time (in seconds since the epoch) of the last time we did a GC for each generation. -_last_gc = [0.0, 0.0, 0.0] - - F = TypeVar("F", bound=Callable[..., Any]) @@ -658,34 +533,6 @@ def runUntilCurrentTimer(reactor: ReactorBase, func: F) -> F: global last_ticked last_ticked = end - if running_on_pypy: - return ret - - # Check if we need to do a manual GC (since its been disabled), and do - # one if necessary. Note we go in reverse order as e.g. a gen 1 GC may - # promote an object into gen 2, and we don't want to handle the same - # object multiple times. - threshold = gc.get_threshold() - counts = gc.get_count() - for i in (2, 1, 0): - # We check if we need to do one based on a straightforward - # comparison between the threshold and count. We also do an extra - # check to make sure that we don't a GC too often. - if threshold[i] < counts[i] and MIN_TIME_BETWEEN_GCS[i] < end - _last_gc[i]: - if i == 0: - logger.debug("Collecting gc %d", i) - else: - logger.info("Collecting gc %d", i) - - start = time.time() - unreachable = gc.collect(i) - end = time.time() - - _last_gc[i] = end - - gc_time.labels(i).observe(end - start) - gc_unreachable.labels(i).set(unreachable) - return ret return cast(F, f) @@ -701,11 +548,6 @@ try: # runUntilCurrent is called when we have pending calls. It is called once # per iteratation after fd polling. reactor.runUntilCurrent = runUntilCurrentTimer(reactor, reactor.runUntilCurrent) # type: ignore - - # We manually run the GC each reactor tick so that we can get some metrics - # about time spent doing GC, - if not running_on_pypy: - gc.disable() except AttributeError: pass @@ -717,4 +559,6 @@ __all__ = [ "LaterGauge", "InFlightGauge", "GaugeBucketCollector", + "MIN_TIME_BETWEEN_GCS", + "install_gc_manager", ] diff --git a/synapse/metrics/_gc.py b/synapse/metrics/_gc.py new file mode 100644 index 000000000..2bc909efa --- /dev/null +++ b/synapse/metrics/_gc.py @@ -0,0 +1,203 @@ +# Copyright 2015-2022 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + + +import gc +import logging +import platform +import time +from typing import Iterable + +from prometheus_client.core import ( + REGISTRY, + CounterMetricFamily, + Gauge, + GaugeMetricFamily, + Histogram, + Metric, +) + +from twisted.internet import task + +"""Prometheus metrics for garbage collection""" + + +logger = logging.getLogger(__name__) + +# The minimum time in seconds between GCs for each generation, regardless of the current GC +# thresholds and counts. +MIN_TIME_BETWEEN_GCS = (1.0, 10.0, 30.0) + +running_on_pypy = platform.python_implementation() == "PyPy" + +# +# Python GC metrics +# + +gc_unreachable = Gauge("python_gc_unreachable_total", "Unreachable GC objects", ["gen"]) +gc_time = Histogram( + "python_gc_time", + "Time taken to GC (sec)", + ["gen"], + buckets=[ + 0.0025, + 0.005, + 0.01, + 0.025, + 0.05, + 0.10, + 0.25, + 0.50, + 1.00, + 2.50, + 5.00, + 7.50, + 15.00, + 30.00, + 45.00, + 60.00, + ], +) + + +class GCCounts: + def collect(self) -> Iterable[Metric]: + cm = GaugeMetricFamily("python_gc_counts", "GC object counts", labels=["gen"]) + for n, m in enumerate(gc.get_count()): + cm.add_metric([str(n)], m) + + yield cm + + +def install_gc_manager() -> None: + """Disable automatic GC, and replace it with a task that runs every 100ms + + This means that (a) we can limit how often GC runs; (b) we can get some metrics + about GC activity. + + It does nothing on PyPy. + """ + + if running_on_pypy: + return + + REGISTRY.register(GCCounts()) + + gc.disable() + + # The time (in seconds since the epoch) of the last time we did a GC for each generation. + _last_gc = [0.0, 0.0, 0.0] + + def _maybe_gc() -> None: + # Check if we need to do a manual GC (since its been disabled), and do + # one if necessary. Note we go in reverse order as e.g. a gen 1 GC may + # promote an object into gen 2, and we don't want to handle the same + # object multiple times. + threshold = gc.get_threshold() + counts = gc.get_count() + end = time.time() + for i in (2, 1, 0): + # We check if we need to do one based on a straightforward + # comparison between the threshold and count. We also do an extra + # check to make sure that we don't a GC too often. + if threshold[i] < counts[i] and MIN_TIME_BETWEEN_GCS[i] < end - _last_gc[i]: + if i == 0: + logger.debug("Collecting gc %d", i) + else: + logger.info("Collecting gc %d", i) + + start = time.time() + unreachable = gc.collect(i) + end = time.time() + + _last_gc[i] = end + + gc_time.labels(i).observe(end - start) + gc_unreachable.labels(i).set(unreachable) + + gc_task = task.LoopingCall(_maybe_gc) + gc_task.start(0.1) + + +# +# PyPy GC / memory metrics +# + + +class PyPyGCStats: + def collect(self) -> Iterable[Metric]: + + # @stats is a pretty-printer object with __str__() returning a nice table, + # plus some fields that contain data from that table. + # unfortunately, fields are pretty-printed themselves (i. e. '4.5MB'). + stats = gc.get_stats(memory_pressure=False) # type: ignore + # @s contains same fields as @stats, but as actual integers. + s = stats._s # type: ignore + + # also note that field naming is completely braindead + # and only vaguely correlates with the pretty-printed table. + # >>>> gc.get_stats(False) + # Total memory consumed: + # GC used: 8.7MB (peak: 39.0MB) # s.total_gc_memory, s.peak_memory + # in arenas: 3.0MB # s.total_arena_memory + # rawmalloced: 1.7MB # s.total_rawmalloced_memory + # nursery: 4.0MB # s.nursery_size + # raw assembler used: 31.0kB # s.jit_backend_used + # ----------------------------- + # Total: 8.8MB # stats.memory_used_sum + # + # Total memory allocated: + # GC allocated: 38.7MB (peak: 41.1MB) # s.total_allocated_memory, s.peak_allocated_memory + # in arenas: 30.9MB # s.peak_arena_memory + # rawmalloced: 4.1MB # s.peak_rawmalloced_memory + # nursery: 4.0MB # s.nursery_size + # raw assembler allocated: 1.0MB # s.jit_backend_allocated + # ----------------------------- + # Total: 39.7MB # stats.memory_allocated_sum + # + # Total time spent in GC: 0.073 # s.total_gc_time + + pypy_gc_time = CounterMetricFamily( + "pypy_gc_time_seconds_total", + "Total time spent in PyPy GC", + labels=[], + ) + pypy_gc_time.add_metric([], s.total_gc_time / 1000) + yield pypy_gc_time + + pypy_mem = GaugeMetricFamily( + "pypy_memory_bytes", + "Memory tracked by PyPy allocator", + labels=["state", "class", "kind"], + ) + # memory used by JIT assembler + pypy_mem.add_metric(["used", "", "jit"], s.jit_backend_used) + pypy_mem.add_metric(["allocated", "", "jit"], s.jit_backend_allocated) + # memory used by GCed objects + pypy_mem.add_metric(["used", "", "arenas"], s.total_arena_memory) + pypy_mem.add_metric(["allocated", "", "arenas"], s.peak_arena_memory) + pypy_mem.add_metric(["used", "", "rawmalloced"], s.total_rawmalloced_memory) + pypy_mem.add_metric(["allocated", "", "rawmalloced"], s.peak_rawmalloced_memory) + pypy_mem.add_metric(["used", "", "nursery"], s.nursery_size) + pypy_mem.add_metric(["allocated", "", "nursery"], s.nursery_size) + # totals + pypy_mem.add_metric(["used", "totals", "gc"], s.total_gc_memory) + pypy_mem.add_metric(["allocated", "totals", "gc"], s.total_allocated_memory) + pypy_mem.add_metric(["used", "totals", "gc_peak"], s.peak_memory) + pypy_mem.add_metric(["allocated", "totals", "gc_peak"], s.peak_allocated_memory) + yield pypy_mem + + +if running_on_pypy: + REGISTRY.register(PyPyGCStats())