This program consumes from the Fastly Real-time Analytics API and makes the data available to Prometheus. It should behave like you expect: dynamically adding new services, removing old services, and reflecting changes to service metadata like name and version.
Go to the releases page.
Available on the packages page as fastly/fastly-exporter.
docker pull ghcr.io/fastly/fastly-exporter:latest
Note that version
latest will track RCs, alphas, etc. -- always use an
explicit version in production.
Helm must be installed to use the prometheus-community/fastly-exporter chart. Please refer to Helm's documentation to get started.
Once Helm is set up properly, add the repo as follows:
helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm upgrade --install fastly-exporter prometheus-fastly-exporter --namespace monitoring --set token="fastly_api_token"
If you have a working Go installation, you can clone the repo and install the binary from any revision, including HEAD.
git clone email@example.com:fastly/fastly-exporter cd fastly-exporter go build ./cmd/fastly-exporter ./fastly-exporter -h
For simple use cases, all you need is a Fastly API token. See this link
for information on creating API tokens. The token can be provided via the
-token flag or the
FASTLY_API_TOKEN environment variable.
fastly-exporter -token XXX
This will collect real-time stats for all Fastly services visible to your token, and make them available as Prometheus metrics on 127.0.0.1:8080/metrics.
By default, all services available to your token will be exported. You can
specify an explicit set of service IDs to export by using the
flag. (Service IDs are available at the top of your Fastly dashboard.) You
can also include only those services whose name matches a regex by using the
-service-allowlist '^Production' flag, or elide any service whose name matches
a regex by using the
-service-blocklist '.*TEST.*' flag.
For tokens with access to a lot of services, it's possible to "shard" the
services among different fastly-exporter instances by using the
flag. For example, to shard all services between 3 exporters, you would start
each exporter as
fastly-exporter [common flags] -service-shard 1/3 fastly-exporter [common flags] -service-shard 2/3 fastly-exporter [common flags] -service-shard 3/3
By default, all metrics provided by the Fastly real-time stats API are exported
as Prometheus metrics. You can export only those metrics whose name matches a
regex by using the
-metric-allowlist 'bytes_total$' flag, or elide any metric
whose name matches a regex by using the
-metric-blocklist imgopto flag.
All flags that filter services or metrics are repeatable. Repeating the same
flag causes its condition to be combined with OR semantics. For example,
-service A -service B would include both services A and B (but not service C).
-service-blocklist Test -service-blocklist Staging would skip any service
whose name contained Test or Staging.
Different flags (for the same filter target) combine with AND semantics. For
-metric-allowlist 'bytes_total$' -metric-blocklist imgopto would only
export metrics whose names ended in bytes_total, but didn't include imgopto.
Per-service metrics are available via
/metrics?target=<service ID>. Available
services are enumerated as targets on the
/sd endpoint, which is compatible
with the generic HTTP service discovery feature of Prometheus. An
example Prometheus scrape config for the Fastly exporter follows.
scrape_configs: - job_name: fastly-exporter http_sd_configs: - url: http://127.0.0.1:8080/sd relabel_configs: - source_labels: [__address__] target_label: __param_target - source_labels: [__param_target] target_label: service - target_label: __address__ replacement: 127.0.0.1:8080