@RunLLM we are planning to use Apache Pulsar as a replacement for Google Pub/Sub. We are only using Shared or Key_Shared subscription types. The total number of topics in the cluster will not exceed 100 and we are planning to use a single tenant and namespace for all of them. What are the metrics that we should set an alert on for monitoring Apache Pulsar?

For monitoring Apache Pulsar, you should set alerts on key metrics at the broker, topic, subscription, and consumer levels. Important metrics include message rates (

pulsar_rate_in

,

pulsar_rate_out

), throughput (

pulsar_throughput_in

,

pulsar_throughput_out

), backlog size (

pulsar_storage_backlog_size

), storage size (

pulsar_storage_size

), publish/dispatch latency (

pulsar_broker_publish_latency

), and consumer/producer counts. For subscriptions, monitor

pulsar_subscription_back_log

,

pulsar_subscription_msg_rate_redeliver

, and for Key_Shared, consider metrics related to message replay and hash blocking if available. For consumers, track

pulsar_consumer_unacked_messages

and

pulsar_consumer_msg_rate_redeliver

as well. You can visualize and alert on these metrics using Grafana dashboards with Prometheus as the data source. StreamNative and Pulsar provide prebuilt dashboards for brokers, topics, and subscriptions. For a full list of available metrics and their descriptions, see the Pulsar metrics reference, topic metrics, subscription metrics, and consumer metrics. Would you like more detail on recommended alert thresholds or example alerting rules? Sources: - Pulsar metrics reference - Topic metrics - Subscription metrics - Consumer metrics - Use Grafana Dashboard 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.