Migrating Metrics

Overview

This section walks you through migrating metrics from Mimir (or Prometheus) to OpenObserve. You will:

1. Assess what metric sources you currently have
2. Identify the migration path for each source type
3. Update configs to point at OpenObserve
4. Validate that metrics are flowing correctly

OpenObserve supports Prometheus remote write natively and accepts OTLP metrics, so migration typically involves changing an endpoint URL in your collector or Prometheus config.

Step 1: Assess Your Current Metric Sources

Run this PromQL query in Grafana (against your Mimir/Prometheus data source) to see what's active:

count({__name__=~".+"}) by (job)

This gives you a list of jobs (e.g. node_exporter, kubernetes-pods, myapp) and series count for each.

Step 2: Categorize Your Sources

Group your metrics by how they're currently collected:

Source Type	Examples	Migration Path
Prometheus server with remote_write to Mimir	Scraping exporters, K8s service discovery	Update remote_write URL
OTel Collector sending to Mimir	prometheusremotewrite exporter	Switch to otlphttp exporter
Grafana Agent / Alloy	Agent metrics block with remote_write	Update endpoint
Telegraf	outputs.http with prometheusremotewrite	Update output URL
Kubernetes (kube-prometheus-stack)	Prometheus Operator, ServiceMonitors	Update remoteWrite in Helm values
AWS CloudWatch metrics	CloudWatch → OTel Collector or Telegraf	See dedicated guide
Azure Monitor metrics	Azure Event Hub → OTel Collector	See dedicated guide

Step 3: Migrate Each Source

From Prometheus Server

If you're running Prometheus with remote_write to Mimir, update the destination URL:

Current config:

# prometheus.yml
remote_write:
  - url: http://mimir:9009/api/v1/push

Update the url to the OpenObserve remote write endpoint and add your credentials.

You can copy the exact Prometheus remote write configuration from the OpenObserve Data Sources UI

Reload Prometheus after updating (no restart needed):

curl -X POST http://localhost:9090/-/reload

Your Prometheus exporters (node_exporter, cAdvisor, kube-state-metrics, mysqld_exporter, etc.) don't change at all — they expose metrics, and Prometheus scrapes them the same way. Only the remote_write destination changes.

---

From OTel Collector

If you're using the OTel Collector with the prometheusremotewrite exporter to send metrics to Mimir, switch to the otlphttp exporter.

Current config:

exporters:
  prometheusremotewrite:
    endpoint: http://mimir:9009/api/v1/push

Copy the exact updated configuration from the Data Sources UI in OpenObserve.

---

From Grafana Agent / Alloy

Grafana Agent reached EOL on November 1, 2025, replaced by Grafana Alloy.

Current Grafana Agent config:

metrics:
  configs:
    - name: default
      remote_write:
        - url: http://mimir:9009/api/v1/push

Updated config:

logs:
  configs:
    - name: default
      clients:
        - url: http://openobserve:5080/api/{org_id}/ingest/metrics/_json
          basic_auth:
            username: admin@example.com
            password: Complexpass#123

Update the remote_write URL to the OpenObserve endpoint. Copy the exact configuration from the Data Sources UI in OpenObserve.

Or replace with OpenObserve Collector.

---

From Telegraf

Current config:

[[outputs.http]]
  url = "http://mimir:9009/api/v1/write"
  data_format = "prometheusremotewrite"

Update the output URL to the OpenObserve endpoint. Copy the exact configuration from the Data Sources UI in OpenObserve.

Dedicated guide : Telegraf → OpenObserve

---

From Kubernetes (kube-prometheus-stack)

If you deployed Prometheus via the kube-prometheus-stack Helm chart with remoteWrite to Mimir:

Create the secret first:

kubectl create secret generic openobserve-secret \
  --from-literal=username=admin@example.com \
  --from-literal=password=Complexpass#123 \
  -n monitoring

Update your Helm values:

prometheus:
  prometheusSpec:
    remoteWrite:
      - url: http://openobserve:5080/api/default/prometheus/api/v1/write
        basicAuth:
          username:
            name: openobserve-secret
            key: username
          password:
            name: openobserve-secret
            key: password

Apply the changes:

helm upgrade kps1 prometheus-community/kube-prometheus-stack -f values.yaml -n monitoring

---

From AWS CloudWatch

For detailed steps on ingesting AWS CloudWatch metrics into OpenObserve, see the dedicated guide:

Dedicated guide: AWS CloudWatch Metrics → OpenObserve

---

From Azure Monitor

For detailed steps on ingesting Azure Monitor metrics into OpenObserve, see the dedicated guide:

Dedicated guide: Azure Monitor Metrics → OpenObserve

---

Step 4: How to Verify

Check in the UI

1. Open the OpenObserve UI → Metrics in the left sidebar.
2. Verify each job from your source inventory appears.
3. Run a test query — sum(process_cpu_utilization{}) or any counter you know is active.
4. Confirm the values and timestamps match what you'd expect from the source.

OpenObserve Metrics Explorer — verify metrics are flowing after migration

Troubleshooting

• No data: Check collector logs for connection errors or auth failures. Confirm the remote_write or otlphttp exporter URL is correct and reachable.
• Missing labels: Ensure your collector isn't stripping labels. Remote write passes all Prometheus labels through by default.
• Case mismatch in PromQL: OpenObserve label matching is case-sensitive. Confirm label values match exactly as ingested.

PromQL Compatibility

OpenObserve supports PromQL for metrics queries. Common functions — rate(), histogram_quantile(), sum by(), avg_over_time(), and label matchers — work as expected.

Next Steps

• Migrating Traces — migrate your trace sources next
• Migrating Logs — migrate your log sources

---

Back to Overview | Previous: Architecture & Terminology | Next: Migrating Traces