Periscope is a self-contained error aggregator that uses in-memory pub/sub for asynchronous event ingestion. The only required dependency is a relational database. Postgres is the preferred database backend, however SQLite can work for small workloads or for non-production environments.
| Name | Description | Default |
|---|---|---|
API_SECRET_KEY_ADMIN |
The API key used for accessing administration endpoints, e.g. creating new projects, listing alerts. | |
POSTGRES_HOST |
Postgres server hostname or IP. | localhost |
POSTGRES_PORT |
Postgres server port. | 5432 |
POSTGRES_USER |
Postgres username. | pguser |
POSTGRES_PASSWORD |
Password for the Postgres user. | |
POSTGRES_DATABASE |
Name of the Postgres database. | periscope |
POSTGRES_ENABLED |
When true the Postgres database configuration is used. |
false |
Periscope accepts events in compliance with the Sentry data model. You can use any official Sentry SDK to send events. An example using the Go SDK can be found in the integration tests.
⚠️ Sentry is an incredible platform for error monitoring! Periscope only aims to provide a very simplistic version of error aggregation functionality, using a simple architecture that should be easily deployed and operated.
flowchart TD
A[CaptureException] -->|Error Event Flush| B(HTTP Ingestion Endpoint)
B --> C{Compute Fingerprint}
C -->|EventGroup Exists| D[Record Only]
C -->|New EventGroup| E[New Alert]
E --> F[Emit Notification]
- Incoming events are validated and routed to projects.
- Each new event is assigned to an event group, based on the extracted fingerprint, in an in-memory background process.
- Each new event group generates an alert and each alert generates notifications for the preconfigured alert destination channels.