GCP Cloud Monitoring

Connect Google Cloud Monitoring when your usage signal already lives in GCP: GKE CPU, Cloud Storage bytes, Cloud SQL connections, Compute Engine utilization, or a custom metric. Costory queries the Monitoring API live and plots those series next to cost in Advanced Explorer. It does not copy metric data. On Add integration, the GCP tile lists Any BigQuery table and Cloud Monitoring metrics. Choose Cloud Monitoring. For tables, see BigQuery.

Add integration page showing Billing tiles and an External metric section with Datadog, AWS, GCP, Amplitude, Tsuga, and Google Sheets — Add integration External metric tiles, including GCP

Manual setup
Terraform

Open Cloud Monitoring

Go to Integrations, click + Add integration, then under External metric choose GCP. On Choose what to connect, pick Cloud Monitoring. The connect URL is /integration/new/gcp/cloud-monitoring. You need to be an admin.

Grant Monitoring Viewer

Copy the Service account from the connect screen. Grant project-level Monitoring Viewer (roles/monitoring.viewer) so Costory can list and query metrics. The role is read-only. It covers every Monitoring series in that project. You cannot scope it to a single metric.Run the command shown under gcloud, or paste this and replace the placeholders:

gcloud projects add-iam-policy-binding PROJECT_ID \
  --member="serviceAccount:SERVICE_ACCOUNT_EMAIL" \
  --role="roles/monitoring.viewer"

The GCP project ID you enter updates the command on the page.

Enter connection details

Set Connection name (default Cloud Monitoring) and GCP project ID. Click Validate. Costory lists active metrics in the project’s metrics scope and shows a count such as Found 42 active metrics in this project’s metrics scope. Click Save connection.Save connection stays disabled until validation succeeds.

Query in Advanced Explorer

Click Explore your data, switch Classic to Advanced, then + Add Metric. Open integrations and pick the Cloud Monitoring source. Search a metric type. The picker shows the path after the service domain, for example instance/cpu/utilization for compute.googleapis.com/instance/cpu/utilization.See Query after you connect.

Enable the Cloud Monitoring API (monitoring.googleapis.com) on the project before you validate. A disabled API fails with a permission-style error.

Grant the role on a scoping project if you use a metrics scope. That one grant covers every project in the scope. You can then group by resource.label.project_id instead of connecting each project.

Connection fields

Field	Description	Default
Connection name	Label on the Integrations list and in the Explorer picker.	Cloud Monitoring
GCP project ID	Project Costory calls on the Monitoring API. Use the scoping project if you monitor many projects from one scope.	(empty)

What you can query

After you save, active metrics from the connected project appear under integrations. Queries run against Cloud Monitoring. Costory does not store the timeseries.

Common series:

Compute Engine CPU utilization (compute.googleapis.com/instance/cpu/utilization)
GKE container CPU request utilization (kubernetes.io/container/cpu/request_utilization)
Cloud Storage object or byte counts (storage.googleapis.com)
Cloud SQL connections or PostgreSQL insights execution time (cloudsql.googleapis.com)
Custom or Prometheus metrics you already export to Cloud Monitoring

Set Aggregate to SUM, AVERAGE, MAX, or MIN. Optional Group by uses Monitoring labels, written as metric.label.* or resource.label.*. resource.label.project_id is always available.To narrow the series, paste a Cloud Monitoring filter in the condition field. The field is not validated in the UI. Example:

resource.label.cluster_name = "prod-gke"

Costory always constrains the query to the metric you picked. Your filter is combined with that metric type.Use the series in a or shared cost allocation rule after you can query it in Explorer. For formulas such as cost per request, see Unit Economics.

Limits

Topic	Details
Active metrics only	Validation and the picker list metrics with data in roughly the last 25 hours. A project with no recent points fails validate.
Numeric metrics	`INT64` and `DOUBLE` values work. `DISTRIBUTION`, `BOOL`, `STRING`, and `MONEY` are not supported.
Group-by cap	Explorer keeps the 200 largest groups by magnitude.
API billing	Listing descriptors is free. Timeseries reads are billed to your GCP billing account.
No PromQL	Pick a metric and optional filter. There is no free-text PromQL or MQL box.

Troubleshooting

Issue	What to check
Permission denied	Grant `roles/monitoring.viewer` to the Costory service account on the same project ID you entered.
Cloud Monitoring API is not enabled	Enable `monitoring.googleapis.com` on that project, then validate again.
Project not found	Confirm the project ID spelling and that the service account can see the project.
No active metrics found	The project (or scope) has no metrics with data in about the last 25 hours. Generate traffic or connect a project that already reports.
Rate limited	Wait a few minutes and retry. Check the project’s Monitoring API quota if it keeps failing.
Chart is empty	Search the full metric type, change Period, and confirm the metric still has recent data. Classic greys out non-cost metrics.
Metric missing from the picker	Type part of the metric type. The first open is a short page of active metrics, not the full Google catalog.

BigQuery metrics

Query a table instead of the Monitoring API on the same GCP tile.

Metrics Correlation

Query external metrics in Advanced Explorer.

Unit Economics

Divide cloud cost by a Cloud Monitoring series.

GCP Billing

Connect Detailed Usage Cost export for GCP spend.

Get Started

Setup

Understand

Tag & Allocate

Share

Teams

Technical Documentation

Glossary

Connection fields

What you can query

Limits

Troubleshooting

BigQuery metrics

Metrics Correlation

Unit Economics

GCP Billing

​Connection fields

​What you can query

​Limits

​Troubleshooting

​Related pages

BigQuery metrics

Metrics Correlation

Unit Economics

GCP Billing

Connection fields

What you can query

Limits

Troubleshooting

Related pages