EventStoreDB metrics
EventStoreDB Metrics Source
The EventStoreDB Metrics source collects operational and performance metrics exposed by an EventStoreDB instance. It periodically scrapes the EventStoreDB statistics endpoint and converts the collected data into metric events that can be forwarded to downstream observability systems.
This source is typically deployed alongside EventStoreDB nodes or as part of a centralized metrics collection layer in on-prem or self-managed environments.
Collection Model
The source operates by polling the EventStoreDB statistics endpoint over HTTP or HTTPS at a fixed interval. Each scrape retrieves the latest metrics snapshot exposed by EventStoreDB and emits it as metric events.
Because the source is stateless, no historical state is preserved between scrapes.
Endpoint Configuration
endpoint (optional, string)
Defines the HTTP or HTTPS endpoint from which EventStoreDB metrics are scraped.
By default, the source targets the standard EventStoreDB statistics endpoint. This parameter should be adjusted when:
- EventStoreDB is exposed on a non-default port
- A reverse proxy or load balancer is used
- TLS termination occurs outside the EventStoreDB node
Metric Namespacing
default_namespace (optional, string)
Overrides the default namespace applied to all emitted metrics.
Namespacing is useful for:
- Avoiding metric name collisions
- Grouping metrics by service or deployment
- Aligning with organizational naming conventions
If not explicitly set, a default namespace associated with EventStoreDB is applied.
Scrape Interval
scrape_interval_secs (optional, uint)
Controls how frequently metrics are collected, expressed in seconds.
Shorter intervals provide higher-resolution metrics but increase:
- Network usage
- Load on the EventStoreDB stats endpoint
Longer intervals reduce overhead but may miss short-lived spikes or transient conditions.
Proxy Configuration
proxy (optional, object)
Controls how outbound HTTP(S) requests to the EventStoreDB endpoint are routed through a proxy.
This is typically required in environments with:
- Restricted outbound network access
- Centralized egress control
- Mandatory corporate proxy usage
proxy.enabled (optional, bool)
Enables or disables proxy support for outbound requests.
When disabled, all requests bypass proxy configuration entirely.
proxy.http / proxy.https (optional, string)
Defines the proxy endpoint to use for HTTP or HTTPS traffic respectively.
These values must be valid URI strings and are applied based on the protocol of the target endpoint.
proxy.no_proxy (optional, [string])
Specifies a list of hosts or address patterns that should bypass the proxy.
This allows direct connections to internal or trusted endpoints while proxying all other traffic.
Reliability and Delivery Semantics
The EventStoreDB Metrics source provides at-least-once delivery, meaning metrics may be re-emitted in the event of downstream delivery failures.
Because acknowledgements are not supported:
- Metric delivery is considered successful once handed off to the pipeline
- Reliability guarantees must be enforced at the sink or global pipeline level
Operational Considerations
- Ensure the EventStoreDB statistics endpoint is reachable from the worker instance
- Protect the stats endpoint using network controls or TLS where possible
- Monitor scrape interval impact on EventStoreDB under high load
- Align metric namespaces with your organization’s monitoring standards
Common Deployment Scenarios
- Sidecar deployment alongside EventStoreDB nodes
- Daemon-style deployment for host-level metrics aggregation
- Centralized scraping of multiple EventStoreDB instances via controlled network access