Redis Metrics Exporter

micromegas-redis-exporter samples a single Redis server over one persistent connection and sends its metrics to a Micromegas stack through the standard ingestion service — no Prometheus layer involved. Run one exporter per Redis instance (e.g. a sidecar or one Deployment per node).

Quick start

docker run -d --name redis-exporter \
  -e MICROMEGAS_TELEMETRY_URL=http://ingestion:9000 \
  -e MICROMEGAS_INGESTION_API_KEY=... \
  -e MICROMEGAS_REDIS_EXPORTER_REDIS_URL=redis://my-redis:6379 \
  -e MICROMEGAS_PROCESS_PROPERTIES=cluster=eu-west,role=cache \
  marcantoinedesroches/micromegas-redis-exporter:latest

Metric presets

Presets are cumulative; select with --metrics / MICROMEGAS_REDIS_EXPORTER_METRICS:

Preset	Collected
core	One INFO per tick: redis_up, memory, clients, throughput, keyspace hits/misses, replication, persistence, per-db key counts
extended	core + SLOWLOG LEN (redis_slowlog_length)
full (default)	extended + per-command stats (redis_command_* tagged command=) + LATENCY LATEST (redis_latency_* tagged event=)

At the default settings (full, 1s interval), a busy server with many distinct commands and latency events can emit several hundred measures per second per exporter — roughly 50M rows/day. Drop to extended/core or widen --sample-interval-seconds to trade detail for volume. On managed Redis offerings that restrict SLOWLOG/LATENCY via ACLs, extended and/or full will report redis_up=0 on every tick (the exporter fails the whole sample if a required command is denied) — drop to core, or to extended if only LATENCY is restricted, on such deployments.

Every metric carries an instance property (from --target-name, default host:port) plus any user-supplied properties, so several exporters can share one stack:

SELECT time, value FROM measures
WHERE name = 'redis_used_memory_bytes'
  AND property_get(properties, 'instance') = 'my-redis:6379';

Process properties vs metric properties

There are two places to hang a tag, and the right one depends on whether the tag varies:

• MICROMEGAS_PROCESS_PROPERTIES (comma-separated key=value) tags the exporter process. Stored once on its processes row and readable from the process_properties column of measures and log_entries, so it also covers the exporter's own log output. Use it for everything constant — cluster, region, role, namespace, pod. This is the standard variable every micromegas process honors; see Telemetry Sink Configuration for the full format and precedence rules.
• --property / MICROMEGAS_REDIS_EXPORTER_PROPERTIES tags every metric, alongside the instance/command/db/event tags the exporter attaches itself. Costs storage on every measure the exporter emits, so reach for it only when a query needs the tag on the measure row itself rather than through a join on the process.

-- constant tag: read it off the process
SELECT time, value FROM measures
WHERE name = 'redis_used_memory_bytes'
  AND property_get(process_properties, 'cluster') = 'prod';

Configuration

Flag	Env	Default
--redis-url	MICROMEGAS_REDIS_EXPORTER_REDIS_URL	redis://127.0.0.1:6379
—	MICROMEGAS_REDIS_EXPORTER_REDIS_PASSWORD	unset (overrides URL password)
--metrics	MICROMEGAS_REDIS_EXPORTER_METRICS	full
--sample-interval-seconds	MICROMEGAS_REDIS_EXPORTER_SAMPLE_INTERVAL_SECONDS	1
--target-name	MICROMEGAS_REDIS_EXPORTER_TARGET_NAME	host:port from the URL
--property k=v (repeatable)	MICROMEGAS_REDIS_EXPORTER_PROPERTIES (comma-separated)	none
--health-listen-addr	MICROMEGAS_REDIS_EXPORTER_HEALTH_LISTEN_ADDR	off
—	MICROMEGAS_PROCESS_PROPERTIES (comma-separated)	none

Telemetry destination, authentication, and process properties use the standard client contract: MICROMEGAS_TELEMETRY_URL plus either MICROMEGAS_INGESTION_API_KEY or the OIDC client-credentials variables (MICROMEGAS_OIDC_TOKEN_ENDPOINT, MICROMEGAS_OIDC_CLIENT_ID, MICROMEGAS_OIDC_CLIENT_SECRET), and MICROMEGAS_PROCESS_PROPERTIES for process-level tags. MICROMEGAS_PROCESS_PROPERTIES has no flag equivalent: the telemetry guard is built before command-line parsing, and a process's properties are sent once at startup.

Kubernetes

The exporter is Kubernetes-friendly but never requires it. Probes are opt-in: /health (liveness) is always 200 while the process runs; /ready is 200 once sampling has started. A Redis outage does not flip readiness — the exporter stays up to report redis_up=0.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: redis-exporter
spec:
  replicas: 1
  selector:
    matchLabels: { app: redis-exporter }
  template:
    metadata:
      labels: { app: redis-exporter }
    spec:
      containers:
        - name: redis-exporter
          image: marcantoinedesroches/micromegas-redis-exporter:latest
          env:
            - name: MICROMEGAS_TELEMETRY_URL
              value: http://ingestion:9000
            - name: MICROMEGAS_INGESTION_API_KEY
              valueFrom:
                secretKeyRef: { name: micromegas-ingestion, key: api-key }
            - name: MICROMEGAS_REDIS_EXPORTER_REDIS_URL
              value: redis://redis:6379
            - name: MICROMEGAS_REDIS_EXPORTER_REDIS_PASSWORD
              valueFrom:
                secretKeyRef: { name: redis, key: password }
            - name: MICROMEGAS_REDIS_EXPORTER_TARGET_NAME
              value: redis-main
            - name: POD_NAMESPACE
              valueFrom:
                fieldRef: { fieldPath: metadata.namespace }
            - name: POD_NAME
              valueFrom:
                fieldRef: { fieldPath: metadata.name }
            - name: MICROMEGAS_PROCESS_PROPERTIES
              value: cluster=prod,namespace=$(POD_NAMESPACE),pod=$(POD_NAME)
            - name: MICROMEGAS_REDIS_EXPORTER_HEALTH_LISTEN_ADDR
              value: 0.0.0.0:8081
          ports:
            - containerPort: 8081
              name: health
          livenessProbe:
            httpGet: { path: /health, port: health }
          readinessProbe:
            httpGet: { path: /ready, port: health }