Apache Pulsar Integration
Monitor topic metrics, subscription backlogs, broker stats, and BookKeeper ledger health across your Pulsar clusters. Get predictive backlog alerts and AI root cause analysis before slow consumers become incidents.
How It Works
Deploy Pulsar Exporter via Helm
Add the TigerOps Helm chart to your cluster and configure your Pulsar broker service URL. The exporter auto-discovers tenants, namespaces, topics, and subscription groups via the Pulsar Admin API.
Configure Remote Write
Point the exporter to your TigerOps remote-write endpoint. Topic backlog sizes, consumer throughput, broker CPU, and BookKeeper write latency all stream in within minutes.
Set Backlog Thresholds
Define per-subscription backlog SLOs. TigerOps fires predictive alerts when backlog growth rate projects a breach and correlates the spike with producer rate changes.
Correlate Across the Stack
TigerOps links Pulsar backlog events with BookKeeper bookie latency, ZooKeeper ensemble health, and downstream consumer service traces for complete incident context.
What You Get Out of the Box
Subscription Backlog Monitoring
Per-subscription and per-topic message backlog tracking with historical trend analysis. Predictive alerts fire before backlog growth impacts downstream consumers.
Broker & Bookie Health
Track Pulsar broker CPU, memory, and network I/O alongside BookKeeper bookie read/write latency, ledger rollover rates, and journal flush times.
Topic & Namespace Metrics
Per-topic publish rates, storage size, replication backlog, and entry sizes. Namespace-level quota utilization and throughput aggregation across tenants.
Consumer Group Throughput
Monitor consumer dispatch rates, acknowledgement rates, and unacknowledged message counts. Detect slow consumers before they create critical subscription backlogs.
Geo-Replication Lag
Track cross-region replication backlog and latency for Pulsar geo-replicated deployments. Alert on replication lag spikes that signal network partition or remote cluster issues.
AI Root Cause Analysis
When subscription backlogs spike, TigerOps AI correlates BookKeeper bookie slowdowns, broker GC pauses, ZooKeeper write latency, and producer batch errors to pinpoint the cause.
Helm Values for pulsar-exporter
Deploy the TigerOps Pulsar exporter to your Kubernetes cluster with these Helm values.
# TigerOps pulsar-exporter Helm values
# helm repo add tigerops https://charts.atatus.net
# helm install pulsar-exporter tigerops/pulsar-exporter -f values.yaml
pulsarExporter:
# Pulsar broker service URL
brokerServiceUrl: pulsar+ssl://pulsar-broker.pulsar.svc.cluster.local:6651
adminApiUrl: https://pulsar-broker.pulsar.svc.cluster.local:8443
# Authentication (OAuth2 token)
auth:
type: oauth2
issuerUrl: https://auth.example.com
clientId: tigerops-exporter
clientSecretRef:
name: pulsar-oauth-secret
key: client_secret
audience: pulsar
# TLS configuration
tls:
enabled: true
caFile: /etc/pulsar-tls/ca.crt
# Scope monitoring to specific tenants/namespaces
tenants:
- public
- payments
namespaceFilter: ".*" # regex; empty = all
remoteWrite:
endpoint: https://ingest.atatus.net/api/v1/write
bearerToken: "${TIGEROPS_API_KEY}"
scrapeInterval: 15s
# Alert thresholds
alerts:
subscriptionBacklogCritical: 500000 # messages
subscriptionBacklogWarning: 50000
replicationBacklogWarning: 10000
bookieWriteLatencyMs: 50Common Questions
Which Pulsar versions does TigerOps support?
TigerOps supports Apache Pulsar 2.8+ and 3.x via the Pulsar Admin REST API and Prometheus metrics endpoint. StreamNative Cloud and Pulsar clusters running on Kubernetes are supported with automatic service discovery.
How does TigerOps collect BookKeeper ledger metrics?
The TigerOps exporter scrapes the BookKeeper bookie Prometheus endpoint (port 8000 by default) alongside the Pulsar broker metrics endpoint. Ledger rollover counts, journal write latency, and entry log compaction metrics are collected automatically.
Can TigerOps monitor multi-tenant Pulsar clusters?
Yes. TigerOps supports full multi-tenant Pulsar deployments. You can filter monitoring scope by tenant, namespace, or topic pattern, and aggregate backlog and throughput metrics at any hierarchy level.
How are Pulsar Function metrics collected?
TigerOps collects Pulsar Functions worker metrics including function execution latency, error rates, and backlog from the Functions worker REST API. You can set per-function SLOs and alert on processing latency breaches.
Does TigerOps support Pulsar with TLS and OAuth2 authentication?
Yes. The TigerOps Pulsar exporter supports TLS-encrypted connections and OAuth2 token-based authentication for the Pulsar Admin API. Configure the token or client credentials in your Helm values and the exporter handles token refresh automatically.
Stop Discovering Pulsar Backlogs After the Fact
Predictive backlog alerts, BookKeeper health monitoring, and AI root cause analysis. Deploy in 5 minutes.