Application Real-Time Monitoring Service:Query metrics and configure a Grafana data source

PromQL quick reference

The following examples cover common query patterns:

# Instant vector selector
up{job="prometheus"}

# Range vector selector (last 5 minutes)
http_requests_total[5m]

# Rate of increase per second
rate(http_requests_total[5m])

# Total increase over one hour
increase(http_requests_total[1h])

# Aggregate by label
sum by (job) (http_requests_total)

# CPU usage percentage
100 - (avg by (instance) (rate(node_cpu_seconds_total{mode="idle"}[5m])) * 100)

# Memory usage percentage
(1 - (node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes)) * 100

# p95 request latency
histogram_quantile(0.95, sum(rate(http_request_duration_seconds_bucket[5m])) by (le))

For the full PromQL specification, see Prometheus documentation.

Step 1: Get the HTTP API URL

1. Log on to the ARMS console.

2. In the left-side navigation pane, choose Managed Service for Prometheus > Instances.

3. In the top navigation bar, select the region of your Prometheus instance. On the Instances page, find the instance and click Settings in the Actions column.

4. On the Settings tab, copy a public or internal URL from the HTTP API URL (Grafana Read URL) section.

   URL type	When to use
   Internal	Your application runs within the same Alibaba Cloud region. Internal traffic is faster and does not incur internet bandwidth charges.
   Public	Your application runs outside Alibaba Cloud or in a different region.

   

Secure API access with a token

To require authentication for all API requests, click Generate Token on the Settings tab. After you generate a token, every HTTP request must include the token in the Authorization header. Without a valid token, requests are rejected.

Step 2: Send a query

The following examples use the Prometheus HTTP API. Replace the placeholders with your values:

Instant query

Returns the value of a metric at a single point in time.

curl -H "Authorization:<token>" \
  "<http-api-url>/api/v1/query?query=<promql>&time=$(date +%s)&timeout=1000"

The time parameter accepts a Unix timestamp. $(date +%s) returns the current timestamp. You can also pass a specific value such as 1709171676.

Range query

Returns the values of a metric over a time range, at regular intervals.

curl -H "Authorization:<token>" \
  "<http-api-url>/api/v1/query_range?query=<promql>&start=$(date -d '1 hour ago' +%s)&end=$(date +%s)&step=60"

The date -d flag is specific to GNU/Linux. On macOS, use date -v-1H +%s instead.

For a complete API reference, see Prometheus HTTP API documentation.

Before you begin

Make sure you have:

• The HTTP API URL from the ARMS console (see Step 1: Get the HTTP API URL in the previous section)

• The authentication token, if you generated one

• Administrator access to your Grafana instance

The following steps use Grafana v10.0.x. The UI may differ slightly in other versions.

Step 1: Add the Prometheus data source

1. Log on to Grafana as an administrator.

   The default credentials for a new Grafana installation are admin / admin. Change the password on first login.

2. Click the icon in the upper-left corner.

3. In the left-side navigation pane, choose Configuration > Data sources.

4. Click Add data source.

5. Select Prometheus.

6. Configure the data source:

   • Name: Enter a descriptive name, for example, arms-prometheus.

   • Prometheus server URL: Paste the HTTP API URL from the ARMS console.

   

7. *(If you generated a token)* In the Custom HTTP Headers section, click + Add header, set Header to Authorization, and set Value to the token.

8. Click Save & test. A success message confirms that Grafana can reach Managed Service for Prometheus.

Step 2: Verify the data source

1. In the left-side navigation pane, click Dashboards.

2. Click New and select New dashboard.

3. Click + Add visualization and select the data source you created.

4. In the A section, enter a metric name in the Metrics field and press Enter. If a chart appears, the data source is working.

   

Troubleshooting

If Grafana runs in a container (Docker or Kubernetes), localhost refers to the container's own network namespace. Use the actual host IP or host.docker.internal instead.

Known limitations

Managed Service for Prometheus remote write does not support HTTP/2.

Step 1: Get the remote read endpoint

1. Log on to the ARMS console.

2. In the left-side navigation pane, choose Managed Service for Prometheus > Instances.

3. In the top navigation bar, select the region of your Prometheus instance. On the Instances page, find the instance and click Settings in the Actions column.

4. On the Settings tab, copy the public or internal remote read endpoint.

   

Step 2: Configure prometheus.yml

1. Download open source Prometheus from the Prometheus download page.

2. Decompress the installation package.

3. Open prometheus.yml and add the following block at the end of the file. Replace the URL with the remote read endpoint from Step 1.

   Placeholder	Description	Example
   <remote-read-endpoint>	The remote read endpoint from the ARMS console	http://cn-hangzhou.arms.aliyuncs.com:9443/api/v1/read

      remote_read:
        - url: "<remote-read-endpoint>"
          read_recent: true

4. Save the file and start (or restart) Prometheus.

Step 3: Verify remote read

1. Open http://localhost:9090 in a browser.

2. In the search box, enter a metric name that exists in Managed Service for Prometheus and click Execute. If the query returns results, remote read is configured correctly.