> ## Documentation Index
> Fetch the complete documentation index at: https://docs.varo.cloud/llms.txt
> Use this file to discover all available pages before exploring further.
# Monitor Deployments with Logs, Metrics, and Alerts
> View real-time logs, application metrics, and set up alerts for your Varo Cloud deployments to catch issues before they affect users.
Varo Cloud gives you visibility into every deployment through real-time logs, application metrics, and configurable alerts. Whether you are debugging a failed build, tracking latency after a release, or setting up proactive notifications for your on-call rotation, all of these tools are available directly from your dashboard at [varo.cloud](https://varo.cloud) and through the API.
## Deployment logs
Every deployment streams logs in real time as your application builds and starts up. To view logs in the dashboard, navigate to **Deployments**, select the deployment you want to inspect, and open the **Logs** tab. You can filter log output by severity level or search for keywords using the toolbar above the log viewer.
Varo Cloud surfaces three log levels:
* **INFO** — general operational messages, startup events, and request traces
* **WARN** — non-fatal conditions that may require attention
* **ERROR** — failures that need immediate investigation
You can also retrieve the status and metadata for a deployment programmatically — for example, to detect failure conditions in a script or CI/CD pipeline. Use the `GET /deployments/{id}` endpoint and inspect the `status` field:
```bash theme={null}
curl https://inference.varo.cloud/v1/deployments/dep_xyz789 \
-H "Authorization: Bearer YOUR_API_KEY"
```
```json theme={null}
{
"id": "dep_xyz789",
"status": "failed",
"environment": "production",
"created_at": "2024-01-15T10:32:01Z",
"error": "Build failed: dependency resolution error"
}
```
When a deployment fails, the `error` field contains a short description of the failure reason. Full build output is always available in the **Logs** tab of the deployment detail page in your dashboard.
## Application metrics
Varo Cloud automatically collects the following metrics for every active deployment:
* **CPU usage** — percentage of allocated CPU consumed
* **Memory usage** — RAM consumed versus the deployment's memory limit
* **Request rate** — incoming HTTP requests per second
* **Error rate** — percentage of requests returning 5xx responses
* **Response latency** — p50, p95, and p99 response times in milliseconds
To view metrics in the dashboard, go to **Deployments**, select a deployment, and open the **Metrics** tab. Graphs update every 30 seconds and support time-range selection for historical analysis.
Metrics retention depends on your plan: **Starter** plans retain metrics for **7 days**, while **Pro** and **Enterprise** plans retain metrics for **30 days**. For longer retention and advanced querying, connect your Datadog account via the Integrations settings.
## Health checks
Varo Cloud periodically pings your application's health check endpoint to verify that it is responding correctly. If the endpoint fails to respond within the configured timeout, Varo Cloud marks the deployment as unhealthy and triggers any alert rules you have configured.
To configure your health check, go to **Project → Settings → Health Check** and fill in the following fields:
* **Health Check URL** — the path Varo Cloud should GET, for example `/healthz`
* **Interval** — how often to run the check, in seconds (minimum 10 seconds)
* **Timeout** — how long to wait for a response before treating the check as failed, in seconds
Varo Cloud expects a `200 OK` response to consider the deployment healthy. Any other status code or a connection timeout counts as a failed check.
## Alerts
Alerts let you define thresholds on any collected metric and get notified through email or Slack when those thresholds are breached.
In your dashboard, go to **Monitoring → Alerts**.
Click the **New Alert** button in the top-right corner of the Alerts page.
Choose the metric you want to monitor, set a threshold value, and define the evaluation window — the period over which the metric must exceed the threshold before the alert fires. For example, you might alert when average CPU usage exceeds 80% for a sustained 5-minute window.
Choose where Varo Cloud should send the alert: **email** (any address on your team) or **Slack** (requires the Slack integration to be connected). You can add multiple notification channels to a single alert.
Here are two common alert configurations to get you started:
**Metric:** CPU usage\
**Threshold:** greater than 80%\
**Window:** 5 minutes\
Fires when sustained compute pressure risks degrading response times or triggering an out-of-memory condition.
**Metric:** Error rate\
**Threshold:** greater than 1%\
**Window:** 2 minutes\
Fires when a meaningful percentage of requests are returning errors, signalling a likely regression in a recent deployment.
## Uptime and status
Uptime tracking gives you a historical view of your deployment's availability over time. It is available on **Pro** and **Enterprise** plans and shows availability percentages across selectable windows of **7 days**, **30 days**, and **90 days**. You can find your uptime history under **Monitoring → Uptime** in the dashboard.
Uptime is calculated based on health check results: a deployment is considered available for any interval in which its health check endpoint returns a successful response.
For advanced observability — including custom dashboards, anomaly detection, and cross-service correlation — connect Datadog to your Varo Cloud account. See the [Integrations guide](/guides/integrations) for step-by-step setup instructions.