Pulsar Metrics

Pulsar exposes metrics in Prometheus format that can be collected and used for monitoring the health of the cluster.

• ZooKeeper
• BookKeeper
• Broker

Overview

The metrics exposed by Pulsar are in Prometheus format. The types of metrics are:

• Counter: a cumulative metric that represents a single monotonically increasing counter whose value can only increase or be reset to zero on restart.
• Gauge: a gauge is a metric that represents a single numerical value that can arbitrarily go up and down.
• Histogram: a histogram samples observations (usually things like request durations or response sizes) and counts them in configurable buckets. The _bucket suffix is the number of observations within a histogram bucket, configured with parameter {le="<upper inclusive bound>"}. The _count suffix is the number of observations, shown as a time series and behaves like a counter. The _sum suffix is the sum of observed values, also shown as a time series and behaves like a counter. These suffixes are together denoted by _* in this doc.
• Summary: similar to a histogram, a summary samples observations (usually things like request durations and response sizes). While it also provides a total count of observations and a sum of all observed values, it calculates configurable quantiles over a sliding time window.

ZooKeeper

The ZooKeeper metrics are exposed under "/metrics" at port 8000. You can use a different port by configuring the stats_server_port system property.

Server metrics

Name	Type	Description
zookeeper_server_znode_count	Gauge	The number of z-nodes stored.
zookeeper_server_data_size_bytes	Gauge	The total size of all of z-nodes stored.
zookeeper_server_connections	Gauge	The number of currently opened connections.
zookeeper_server_watches_count	Gauge	The number of watchers registered.
zookeeper_server_ephemerals_count	Gauge	The number of ephemeral z-nodes.

Request metrics

Name	Type	Description
zookeeper_server_requests	Counter	The total number of requests received by a particular server.
zookeeper_server_requests_latency_ms	Summary	The requests latency calculated in milliseconds. Available labels: type (write, read). write: the requests that write data to ZooKeeper. read: the requests that read data from ZooKeeper.

BookKeeper

The BookKeeper metrics are exposed under "/metrics" at port 8000. You can change the port by updating prometheusStatsHttpPort in bookkeeper.conf configuration file.

Server metrics

Name	Type	Description
bookie_SERVER_STATUS	Gauge	The server status for bookie server. 1: the bookie is running in writable mode. 0: the bookie is running in readonly mode.
bookkeeper_server_ADD_ENTRY_count	Counter	The total number of ADD_ENTRY requests received at the bookie. The success label is used to distinguish successes and failures.
bookkeeper_server_READ_ENTRY_count	Counter	The total number of READ_ENTRY requests received at the bookie. The success label is used to distinguish successes and failures.
bookie_WRITE_BYTES	Counter	The total number of bytes written to the bookie.
bookie_READ_BYTES	Counter	The total number of bytes read from the bookie.
bookkeeper_server_ADD_ENTRY_REQUEST	Summary	The summary of request latency of ADD_ENTRY requests at the bookie. The success label is used to distinguish successes and failures.
bookkeeper_server_READ_ENTRY_REQUEST	Summary	The summary of request latency of READ_ENTRY requests at the bookie. The success label is used to distinguish successes and failures.

Journal metrics

Name	Type	Description
bookie_journal_JOURNAL_SYNC_count	Counter	The total number of journal fsync operations happening at the bookie. The success label is used to distinguish successes and failures.
bookie_journal_JOURNAL_QUEUE_SIZE	Gauge	The total number of requests pending in the journal queue.
bookie_journal_JOURNAL_FORCE_WRITE_QUEUE_SIZE	Gauge	The total number of force write (fsync) requests pending in the force-write queue.
bookie_journal_JOURNAL_CB_QUEUE_SIZE	Gauge	The total number of callbacks pending in the callback queue.
bookie_journal_JOURNAL_ADD_ENTRY	Summary	The summary of request latency of adding entries to the journal.
bookie_journal_JOURNAL_SYNC	Summary	The summary of fsync latency of syncing data to the journal disk.

Storage metrics

Name	Type	Description
bookie_ledgers_count	Gauge	The total number of ledgers stored in the bookie.
bookie_entries_count	Gauge	The total number of entries stored in the bookie.
bookie_write_cache_size	Gauge	The bookie write cache size (in bytes).
bookie_read_cache_size	Gauge	The bookie read cache size (in bytes).
bookie_DELETED_LEDGER_COUNT	Counter	The total number of ledgers deleted since the bookie has started.
bookie_ledger_writable_dirs	Gauge	The number of writable directories in the bookie.

Broker

The broker metrics are exposed under "/metrics/" at port 8080. You can change the port by updating webServicePort to a different port in broker.conf configuration file.

All the metrics exposed by a broker are labelled with cluster=\$\{pulsar_cluster\}. The value of \$\{pulsar_cluster\} is the pulsar cluster name you configured in broker.conf.

Broker has the following kinds of metrics:

• Namespace metrics
  • Replication metrics
• Topic metrics
  • Replication metrics
• Subscription metrics
• Consumer metrics
• Jetty metrics

Namespace metrics

Namespace metrics are only exposed when exposeTopicLevelMetricsInPrometheus is set to false.

All the namespace metrics are labelled with the following labels:

• cluster: cluster=\$\{pulsar_cluster\}. \$\{pulsar_cluster\} is the cluster name that you configured in broker.conf.
• namespace: namespace=\$\{pulsar_namespace\}. \$\{pulsar_namespace\} is the namespace name.

Name	Type	Description
pulsar_topics_count	Gauge	The number of Pulsar topics of the namespace owned by this broker.
pulsar_subscriptions_count	Gauge	The number of Pulsar subscriptions of the namespace served by this broker.
pulsar_producers_count	Gauge	The number of active producers of the namespace connected to this broker.
pulsar_consumers_count	Gauge	The number of active consumers of the namespace connected to this broker.
pulsar_rate_in	Gauge	The total message rate of the namespace coming into this broker (messages/second).
pulsar_rate_out	Gauge	The total message rate of the namespace going out from this broker (messages/second).
pulsar_throughput_in	Gauge	The total throughput of the namespace coming into this broker (bytes/second).
pulsar_throughput_out	Gauge	The total throughput of the namespace going out from this broker (bytes/second).
pulsar_storage_size	Gauge	The total storage size of the topics in this namespace owned by this broker (bytes).
pulsar_storage_backlog_size	Gauge	The total backlog size of the topics of this namespace owned by this broker (messages).
pulsar_storage_offloaded_size	Gauge	The total amount of the data in this namespace offloaded to the tiered storage (bytes).
pulsar_storage_write_rate	Gauge	The total message batches (entries) written to the storage for this namespace (message batches / second).
pulsar_storage_read_rate	Gauge	The total message batches (entries) read from the storage for this namespace (message batches / second).
pulsar_subscription_delayed	Gauge	The total message batches (entries) are delayed for dispatching.
pulsar_storage_write_latency_le_*	Histogram	The entry rate of a namespace that the storage write latency is smaller with a given threshold. Available thresholds: pulsar_storage_write_latency_le_0_5: <= 0.5ms pulsar_storage_write_latency_le_1: <= 1ms pulsar_storage_write_latency_le_5: <= 5ms pulsar_storage_write_latency_le_10: <= 10ms pulsar_storage_write_latency_le_20: <= 20ms pulsar_storage_write_latency_le_50: <= 50ms pulsar_storage_write_latency_le_100: <= 100ms pulsar_storage_write_latency_le_200: <= 200ms pulsar_storage_write_latency_le_1000: <= 1s pulsar_storage_write_latency_le_overflow: > 1s
pulsar_entry_size_le_*	Histogram	The entry rate of a namespace that the entry size is smaller with a given threshold. Available thresholds: pulsar_entry_size_le_128: <= 128 bytes pulsar_entry_size_le_512: <= 512 bytes pulsar_entry_size_le_1_kb: <= 1 KB pulsar_entry_size_le_2_kb: <= 2 KB pulsar_entry_size_le_4_kb: <= 4 KB pulsar_entry_size_le_16_kb: <= 16 KB pulsar_entry_size_le_100_kb: <= 100 KB pulsar_entry_size_le_1_mb: <= 1 MB pulsar_entry_size_le_overflow: > 1 MB

Replication metrics

If a namespace is configured to be replicated between multiple Pulsar clusters, the corresponding replication metrics will also be exposed when replicationMetricsEnabled is enabled.

All the replication metrics will also be labelled with remoteCluster=\$\{pulsar_remote_cluster\}.

Name	Type	Description
pulsar_replication_rate_in	Gauge	The total message rate of the namespace replicating from remote cluster (messages/second).
pulsar_replication_rate_out	Gauge	The total message rate of the namespace replicating to remote cluster (messages/second).
pulsar_replication_throughput_in	Gauge	The total throughput of the namespace replicating from remote cluster (bytes/second).
pulsar_replication_throughput_out	Gauge	The total throughput of the namespace replicating to remote cluster (bytes/second).
pulsar_replication_backlog	Gauge	The total backlog of the namespace replicating to remote cluster (messages).

Topic metrics

Topic metrics are only exposed when exposeTopicLevelMetricsInPrometheus is set to true.

All the topic metrics are labelled with the following labels:

• cluster: cluster=\$\{pulsar_cluster\}. \$\{pulsar_cluster\} is the cluster name that you configured in broker.conf.
• namespace: namespace=\$\{pulsar_namespace\}. \$\{pulsar_namespace\} is the namespace name.
• topic: topic=\$\{pulsar_topic\}. \$\{pulsar_topic\} is the topic name.

Name	Type	Description
pulsar_subscriptions_count	Gauge	The number of Pulsar subscriptions of the topic served by this broker.
pulsar_producers_count	Gauge	The number of active producers of the topic connected to this broker.
pulsar_consumers_count	Gauge	The number of active consumers of the topic connected to this broker.
pulsar_rate_in	Gauge	The total message rate of the topic coming into this broker (messages/second).
pulsar_rate_out	Gauge	The total message rate of the topic going out from this broker (messages/second).
pulsar_throughput_in	Gauge	The total throughput of the topic coming into this broker (bytes/second).
pulsar_throughput_out	Gauge	The total throughput of the topic going out from this broker (bytes/second).
pulsar_storage_size	Gauge	The total storage size of the topics in this topic owned by this broker (bytes).
pulsar_storage_backlog_size	Gauge	The total backlog size of the topics of this topic owned by this broker (messages).
pulsar_storage_offloaded_size	Gauge	The total amount of the data in this topic offloaded to the tiered storage (bytes).
pulsar_storage_write_rate	Gauge	The total message batches (entries) written to the storage for this topic (message batches / second).
pulsar_storage_read_rate	Gauge	The total message batches (entries) read from the storage for this topic (message batches / second).
pulsar_subscription_delayed	Gauge	The total message batches (entries) are delayed for dispatching.
pulsar_storage_write_latency_le_*	Histogram	The entry rate of a topic that the storage write latency is smaller with a given threshold. Available thresholds: pulsar_storage_write_latency_le_0_5: <= 0.5ms pulsar_storage_write_latency_le_1: <= 1ms pulsar_storage_write_latency_le_5: <= 5ms pulsar_storage_write_latency_le_10: <= 10ms pulsar_storage_write_latency_le_20: <= 20ms pulsar_storage_write_latency_le_50: <= 50ms pulsar_storage_write_latency_le_100: <= 100ms pulsar_storage_write_latency_le_200: <= 200ms pulsar_storage_write_latency_le_1000: <= 1s pulsar_storage_write_latency_le_overflow: > 1s
pulsar_entry_size_le_*	Histogram	The entry rate of a topic that the entry size is smaller with a given threshold. Available thresholds: pulsar_entry_size_le_128: <= 128 bytes pulsar_entry_size_le_512: <= 512 bytes pulsar_entry_size_le_1_kb: <= 1 KB pulsar_entry_size_le_2_kb: <= 2 KB pulsar_entry_size_le_4_kb: <= 4 KB pulsar_entry_size_le_16_kb: <= 16 KB pulsar_entry_size_le_100_kb: <= 100 KB pulsar_entry_size_le_1_mb: <= 1 MB pulsar_entry_size_le_overflow: > 1 MB
pulsar_in_bytes_total	Counter	The total number of bytes received for this topic
pulsar_producers_count	Counter	The total number of messages received for this topic

Replication metrics

If a namespace that a topic belongs to is configured to be replicated between multiple Pulsar clusters, the corresponding replication metrics will also be exposed when replicationMetricsEnabled is enabled.

All the replication metrics will also be labelled with remoteCluster=\$\{pulsar_remote_cluster\}.

Name	Type	Description
pulsar_replication_rate_in	Gauge	The total message rate of the topic replicating from remote cluster (messages/second).
pulsar_replication_rate_out	Gauge	The total message rate of the topic replicating to remote cluster (messages/second).
pulsar_replication_throughput_in	Gauge	The total throughput of the topic replicating from remote cluster (bytes/second).
pulsar_replication_throughput_out	Gauge	The total throughput of the topic replicating to remote cluster (bytes/second).
pulsar_replication_backlog	Gauge	The total backlog of the topic replicating to remote cluster (messages).

Subscription metrics

Subscription metrics are only exposed when exposeTopicLevelMetricsInPrometheus is set to true.

All the subscription metrics are labelled with the following labels:

• cluster: cluster=\$\{pulsar_cluster\}. \$\{pulsar_cluster\} is the cluster name that you configured in broker.conf.
• namespace: namespace=\$\{pulsar_namespace\}. \$\{pulsar_namespace\} is the namespace name.
• topic: topic=\$\{pulsar_topic\}. \$\{pulsar_topic\} is the topic name.
• subscription: subscription=\$\{subscription\}. \$\{subscription\} is the topic subscription name.

Name	Type	Description
pulsar_subscription_back_log	Gauge	The total backlog of a subscription (messages).
pulsar_subscription_delayed	Gauge	The total number of messages are delayed to be dispatched for a subscription (messages).
pulsar_subscription_msg_rate_redeliver	Gauge	The total message rate for message being redelivered (messages/second).
pulsar_subscription_unacked_messages	Gauge	The total number of unacknowledged messages of a subscription (messages).
pulsar_subscription_blocked_on_unacked_messages	Gauge	Indicate whether a subscription is blocked on unacknowledged messages or not. 1 means the subscription is blocked on waiting unacknowledged messages to be acked. 0 means the subscription is not blocked on waiting unacknowledged messages to be acked.
pulsar_subscription_msg_rate_out	Gauge	The total message dispatch rate for a subscription (messages/second).
pulsar_subscription_msg_throughput_out	Gauge	The total message dispatch throughput for a subscription (bytes/second).

Consumer metrics

Consumer metrics are only exposed when both exposeTopicLevelMetricsInPrometheus and exposeConsumerLevelMetricsInPrometheus are set to true.

All the consumer metrics are labelled with the following labels:

• cluster: cluster=\$\{pulsar_cluster\}. \$\{pulsar_cluster\} is the cluster name that you configured in broker.conf.
• namespace: namespace=\$\{pulsar_namespace\}. \$\{pulsar_namespace\} is the namespace name.
• topic: topic=\$\{pulsar_topic\}. \$\{pulsar_topic\} is the topic name.
• subscription: subscription=\$\{subscription\}. \$\{subscription\} is the topic subscription name.
• consumer_name: consumer_name=\$\{consumer_name\}. \$\{consumer_name\} is the topic consumer name.
• consumer_id: consumer_id=\$\{consumer_id\}. \$\{consumer_id\} is the topic consumer id.

Name	Type	Description
pulsar_consumer_msg_rate_redeliver	Gauge	The total message rate for message being redelivered (messages/second).
pulsar_consumer_unacked_messages	Gauge	The total number of unacknowledged messages of a consumer (messages).
pulsar_consumer_blocked_on_unacked_messages	Gauge	Indicate whether a consumer is blocked on unacknowledged messages or not. 1 means the consumer is blocked on waiting unacknowledged messages to be acked. 0 means the consumer is not blocked on waiting unacknowledged messages to be acked.
pulsar_consumer_msg_rate_out	Gauge	The total message dispatch rate for a consumer (messages/second).
pulsar_consumer_msg_throughput_out	Gauge	The total message dispatch throughput for a consumer (bytes/second).
pulsar_consumer_available_permits	Gauge	The available permits for for a consumer.

Jetty metrics

For a functions-worker running separately from brokers, its Jetty metrics are only exposed when includeStandardPrometheusMetrics is set to true.

All the jetty metrics are labelled with the following labels:

• cluster: cluster=\$\{pulsar_cluster\}. \$\{pulsar_cluster\} is the cluster name that you have configured in the broker.conf file.

Name	Type	Description
jetty_requests_total	Counter	Number of requests.
jetty_requests_active	Gauge	Number of requests currently active.
jetty_requests_active_max	Gauge	Maximum number of requests that have been active at once.
jetty_request_time_max_seconds	Gauge	Maximum time spent handling requests.
jetty_request_time_seconds_total	Counter	Total time spent in all request handling.
jetty_dispatched_total	Counter	Number of dispatches.
jetty_dispatched_active	Gauge	Number of dispatches currently active.
jetty_dispatched_active_max	Gauge	Maximum number of active dispatches being handled.
jetty_dispatched_time_max	Gauge	Maximum time spent in dispatch handling.
jetty_dispatched_time_seconds_total	Counter	Total time spent in dispatch handling.
jetty_async_requests_total	Counter	Total number of async requests.
jetty_async_requests_waiting	Gauge	Currently waiting async requests.
jetty_async_requests_waiting_max	Gauge	Maximum number of waiting async requests.
jetty_async_dispatches_total	Counter	Number of requested that have been asynchronously dispatched.
jetty_expires_total	Counter	Number of async requests requests that have expired.
jetty_responses_total	Counter	Number of responses, labeled by status code. The code label can be "1xx", "2xx", "3xx", "4xx", or "5xx".
jetty_stats_seconds	Gauge	Time in seconds stats have been collected for.
jetty_responses_bytes_total	Counter	Total number of bytes across all responses.

Monitor

You can set up a Prometheus instance to collect all the metrics exposed at Pulsar components and set up Grafana dashboards to display the metrics and monitor your Pulsar cluster.

The following are some Grafana dashboards examples:

• pulsar-grafana: A grafana dashboard that displays metrics collected in Prometheus for Pulsar clusters running on Kubernetes.
• apache-pulsar-grafana-dashboard: A collection of grafana dashboard templates for different Pulsar components running on both Kubernetes and on-premise machines.