Start a metric from a template
Pre-filled metric definitions for common patterns. Pick one, fill a couple of parameters, edit anything, save.
A metric template is a pre-filled definition for a common monitoring pattern: an HTTP error rate, an RDS CPU metric, a Postgres replication lag query, a TLS expiry check. Instead of starting from a blank form and typing the right query and thresholds yourself, you pick a template, fill in the few values specific to your environment, and save.
Templates are a starting point, not a constraint. Every field they pre-fill stays editable before you save.
Using a template
- Go to Metrics, then New metric.
- In the Source section, choose Start from a template.
- Browse by category or search by name. Each card shows what the template monitors and which source type it uses.
- Pick a template and fill its parameters (for example the RDS instance ID, or the env var holding a database connection string).
- Apply. The form is now pre-filled: source type, configuration, suggested thresholds, and a suggested name and description.
- Adjust anything, then save like any other metric.
What a template fills in
- Source type and configuration, with your parameter values substituted in.
- Suggested thresholds tuned for what the metric measures (for example "under 80 healthy, over 95 unhealthy" for CPU).
- A suggested name and description, used only if you haven't already typed your own.
Parameters
A template pre-fills structure; the values unique to your environment are parameters you fill in. A parameter is a hostname, an AWS region, an RDS instance ID, or the name of an environment variable on the agent that holds a secret. Parameters marked with an asterisk are required.
For sources that read a secret (databases, authenticated log backends), the parameter is the name of an environment variable on the agent, never the secret itself. The template carries the query; you supply the reference.
Available templates
| Category | Templates |
|---|---|
| Metrics | HTTP request rate, HTTP 5xx error rate, HTTP p95 latency, Node CPU saturation (Prometheus) |
| Cloud | RDS CPU, RDS connections, Lambda error count, ALB 5xx count (CloudWatch) |
| Database | Postgres replication lag, Postgres connections, MySQL connections |
| Logs | Application error rate (Loki), Error log count (Elasticsearch) |
| Network | TCP port reachable, DNS resolution, TLS certificate expiry, Host reachable (ICMP) |
| Web | GitHub API rate limit remaining, HTTP endpoint health |
| Realtime | WebSocket connection success, gRPC health check |
After applying
The pre-filled values are suggestions. Review the query, confirm the thresholds match your environment, point the metric at the right agent, and set the polling interval. Save when it reads correctly. If a value looks wrong, change it: the template has done its job once the form is populated.
Suggesting a template
The template set is curated. If a pattern you reach for often isn't here, ask your Observer contact to add it. New templates are a small change and ship in a regular release.