Grafana Cloud (Mimir)¶
This guide walks you through integrating Robusta with Grafana Cloud, enabling both the Robusta runner and Holmes to query metrics from Mimir and receive alerts from Grafana Cloud AlertManager.
Prerequisites¶
Before starting, ensure you have:
A Grafana Cloud account with a configured instance
Prometheus and AlertManager datasources configured in Grafana Cloud
Access to create service accounts and API tokens in Grafana Cloud
Your Robusta
account_id
andsigning_key
fromgenerated_values.yaml
Step 1: Gather Grafana Cloud Information¶
Find Your Grafana Instance Details¶
Log into your Grafana Cloud portal
Note your Grafana instance URL (e.g.,
https://YOUR-INSTANCE.grafana.net
)
Create API Keys¶
You'll need credentials for Grafana API Access (used by both Robusta Runner and Holmes):
Go to Administration → Service accounts
Create a new service account named "robusta-integration"
Generate a service account token
Save the token (starts with
glsa_
)
Find Your Cluster Name¶
If your grafana setup covers multiple clusters, the cluster name is required and used to identify your specific cluster in Prometheus queries:
Go to Grafana → Explore
Run query:
up{cluster!=""}
Check the cluster label values
This value will be set to
cluster_name
in yourgenerated_values.yaml
Find Datasource UIDs¶
Using the Grafana API, list your datasources:
curl -H "Authorization: Bearer YOUR_GLSA_TOKEN" \
"https://YOUR-INSTANCE.grafana.net/api/datasources" | jq
Note the UID for Prometheus datasource UID (typically grafanacloud-prom
)
Step 2: Configure Robusta Runner¶
Update Robusta Values¶
Add the following to your generated_values.yaml
:
globalConfig:
# Your Robusta account details (should already exist)
account_id: YOUR_ROBUSTA_ACCOUNT_ID
signing_key: YOUR_ROBUSTA_SIGNING_KEY
# Grafana Cloud Prometheus Configuration (via proxy)
prometheus_url: https://YOUR-INSTANCE.grafana.net/api/datasources/proxy/uid/PROMETHEUS_DATASOURCE_UID
prometheus_auth: Bearer YOUR_GLSA_TOKEN
# Grafana Cloud AlertManager Configuration
alertmanager_url: https://YOUR-INSTANCE.grafana.net
alertmanager_flavor: grafana
alertmanager_auth: Bearer YOUR_GLSA_TOKEN
# Grafana API Key for enrichments
grafana_api_key: YOUR_GLSA_TOKEN
# Cluster identification (required)
cluster_name: YOUR_CLUSTER_NAME
Note
The prometheus_url
uses Grafana's proxy endpoint format which handles authentication and routing to Mimir automatically.
Apply Configuration¶
Apply the configuration changes:
helm upgrade robusta robusta/robusta -f generated_values.yaml -n default
Restart Robusta Runner¶
Ensure the changes take effect:
kubectl rollout restart deployment/robusta-runner -n default
Step 3: Configure Holmes Prometheus Toolset¶
Holmes requires additional configuration to work with Grafana Cloud's Mimir backend.
For detailed instructions on configuring Holmes with Grafana Cloud, see the Grafana Cloud (Mimir) Configuration section in Prometheus.
The key configuration points for Grafana Cloud are:
Use the proxy endpoint URL format:
https://YOUR-INSTANCE.grafana.net/api/datasources/proxy/uid/PROMETHEUS_DATASOURCE_UID
Set
fetch_labels_with_labels_api: false
(important for Mimir compatibility)Set
fetch_metadata_with_series_api: true
(important for Mimir compatibility)Use Bearer authentication with your service account token
After updating your generated_values.yaml
with the Holmes configuration, apply the changes:
helm upgrade robusta robusta/robusta -f generated_values.yaml -n default
kubectl rollout restart deployment/robusta-holmes -n default
Step 4: Configure Alert Routing (Optional)¶
To send alerts from Grafana Cloud to Robusta's timeline, follow the alert configuration steps in Grafana AlertManager.
The key differences for Grafana Cloud are:
Use your Grafana Cloud instance URL
Use the service account token (
glsa_
token) for authenticationEnsure your alerts include the
cluster
label matching your configuredcluster_name
Verification¶
Verify Metrics Integration¶
Open any application in the Robusta UI
Check if CPU and memory graphs are displayed
If graphs are shown, the metrics integration is working correctly
Verify Holmes Integration¶
Trigger a test alert or wait for an actual alert
In the Robusta UI, click on "Investigate with Holmes"
Verify that Holmes can query metrics and provide analysis
Troubleshooting¶
Common Issues¶
Metrics not showing in Robusta UI:
Verify the
prometheus_url
includes the correct datasource UIDCheck that the service account token has not expired
Ensure the token has appropriate permissions to query metrics
Holmes unable to query metrics:
Verify
fetch_metadata_with_series_api
is set totrue
Check that the Holmes deployment has restarted after configuration changes
Review Holmes logs for authentication errors:
kubectl logs -n default deployment/robusta-holmes
Authentication errors:
Regenerate the service account token if expired
Ensure the token is correctly formatted with ``Bearer `` prefix
Verify the token has the necessary permissions in Grafana Cloud
Additional Resources¶
Grafana AlertManager - For configuring Grafana alerts
Prometheus - For advanced Holmes configuration