asafm commented on code in PR #22058:
URL: https://github.com/apache/pulsar/pull/22058#discussion_r1507304541
##########
pulsar-broker/src/main/java/org/apache/pulsar/broker/service/BrokerService.java:
##########
@@ -403,15 +412,43 @@ public BrokerService(PulsarService pulsar, EventLoopGroup
eventLoopGroup) throws
this.defaultServerBootstrap = defaultServerBootstrap();
this.pendingLookupRequests =
ObserverGauge.build("pulsar_broker_lookup_pending_requests", "-")
- .supplier(() ->
pulsar.getConfig().getMaxConcurrentLookupRequest()
- - lookupRequestSemaphore.get().availablePermits())
+ .supplier(this::getPendingLookupRequest)
.register();
+ this.pendingLookupOperationsCounter =
pulsar.getOpenTelemetry().getMeter()
+
.upDownCounterBuilder("pulsar.broker.topic.lookup.operation.pending.usage")
+ .setDescription("The number of pending lookup operations in
the broker. "
+ + "When it reaches threshold
\"maxConcurrentLookupRequest\" defined in broker.conf, "
+ + "new requests are rejected.")
+ .setUnit("{operation}")
+ .buildWithCallback(measurement ->
measurement.record(getPendingLookupRequest()));
+ this.pendingLookupOperationsLimitGauge =
pulsar.getOpenTelemetry().getMeter()
+
.gaugeBuilder("pulsar.broker.topic.lookup.operation.pending.limit")
Review Comment:
I can presume that people may want a cluster-wide sum of topic lookup
operations since they see that ZK is under stress and would like to know the
cause, which a single broker does not cause. Once you define the pending usage
as UpDownCounter, it only makes sense the limit is also UpDownCounter.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
