From 3baf6e1667f0c257f288a28bd542a93c2445fd30 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 21 Sep 2018 16:25:42 +0100 Subject: Fix ExpiringCache.__len__ to be accurate It used to try and produce an estimate, which was sometimes negative. This caused metrics to be sad, so lets always just calculate it from scratch. (This appears to have been a longstanding bug, but one which has been made more of a problem by #3932 and #3933). (This was originally done by Erik as part of #3933. I'm cherry-picking it because really it's a fix in its own right) --- synapse/util/caches/expiringcache.py | 17 +++++++---------- 1 file changed, 7 insertions(+), 10 deletions(-) (limited to 'synapse/util') diff --git a/synapse/util/caches/expiringcache.py b/synapse/util/caches/expiringcache.py index 921a9c5b29..9af4ec4aa8 100644 --- a/synapse/util/caches/expiringcache.py +++ b/synapse/util/caches/expiringcache.py @@ -16,6 +16,8 @@ import logging from collections import OrderedDict +from six import itervalues + from synapse.metrics.background_process_metrics import run_as_background_process from synapse.util.caches import register_cache @@ -54,8 +56,6 @@ class ExpiringCache(object): self.iterable = iterable - self._size_estimate = 0 - self.metrics = register_cache("expiring", cache_name, self) if not self._expiry_ms: @@ -74,16 +74,11 @@ class ExpiringCache(object): now = self._clock.time_msec() self._cache[key] = _CacheEntry(now, value) - if self.iterable: - self._size_estimate += len(value) - # Evict if there are now too many items while self._max_len and len(self) > self._max_len: _key, value = self._cache.popitem(last=False) if self.iterable: - removed_len = len(value.value) - self.metrics.inc_evictions(removed_len) - self._size_estimate -= removed_len + self.metrics.inc_evictions(len(value.value)) else: self.metrics.inc_evictions() @@ -134,7 +129,9 @@ class ExpiringCache(object): for k in keys_to_delete: value = self._cache.pop(k) if self.iterable: - self._size_estimate -= len(value.value) + self.metrics.inc_evictions(len(value.value)) + else: + self.metrics.inc_evictions() logger.debug( "[%s] _prune_cache before: %d, after len: %d", @@ -143,7 +140,7 @@ class ExpiringCache(object): def __len__(self): if self.iterable: - return self._size_estimate + return sum(len(entry.value) for entry in itervalues(self._cache)) else: return len(self._cache) -- cgit 1.4.1 From 7ee94fc1baf6c4e2970232383d76277d8378712b Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Wed, 26 Sep 2018 12:40:20 +0100 Subject: Log which cache is throwing exceptions --- synapse/util/caches/__init__.py | 27 +++++++++++++++++---------- 1 file changed, 17 insertions(+), 10 deletions(-) (limited to 'synapse/util') diff --git a/synapse/util/caches/__init__.py b/synapse/util/caches/__init__.py index 7b065b195e..f37d5bec08 100644 --- a/synapse/util/caches/__init__.py +++ b/synapse/util/caches/__init__.py @@ -13,6 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging import os import six @@ -20,6 +21,8 @@ from six.moves import intern from prometheus_client.core import REGISTRY, Gauge, GaugeMetricFamily +logger = logging.getLogger(__name__) + CACHE_SIZE_FACTOR = float(os.environ.get("SYNAPSE_CACHE_FACTOR", 0.5)) @@ -76,16 +79,20 @@ def register_cache(cache_type, cache_name, cache): return [] def collect(self): - if cache_type == "response_cache": - response_cache_size.labels(cache_name).set(len(cache)) - response_cache_hits.labels(cache_name).set(self.hits) - response_cache_evicted.labels(cache_name).set(self.evicted_size) - response_cache_total.labels(cache_name).set(self.hits + self.misses) - else: - cache_size.labels(cache_name).set(len(cache)) - cache_hits.labels(cache_name).set(self.hits) - cache_evicted.labels(cache_name).set(self.evicted_size) - cache_total.labels(cache_name).set(self.hits + self.misses) + try: + if cache_type == "response_cache": + response_cache_size.labels(cache_name).set(len(cache)) + response_cache_hits.labels(cache_name).set(self.hits) + response_cache_evicted.labels(cache_name).set(self.evicted_size) + response_cache_total.labels(cache_name).set(self.hits + self.misses) + else: + cache_size.labels(cache_name).set(len(cache)) + cache_hits.labels(cache_name).set(self.hits) + cache_evicted.labels(cache_name).set(self.evicted_size) + cache_total.labels(cache_name).set(self.hits + self.misses) + except Exception as e: + logger.warn("Error calculating metrics for %s: %s", cache_name, e) + raise yield GaugeMetricFamily("__unused", "") -- cgit 1.4.1