v7.0.0-beta.1

This is a pre-release which includes several significant changes, including one breaking change — the -endpoint http://localhost:8080/metrics flag is removed and replaced by -listen localhost:8080, which means the /metrics endpoint is no longer configurable. The exporter now has a human-friendly-ish index page at the default / route. Finally, "official" Docker images are now available as ghcr.io/peterbourgon/fastly-exporter and visible on the GitHub packages page.

Changelog highlights:

• Support for new Compute@Edge metrics — #72
• Refactor for Prometheus HTTP service discovery — #73
• Update README with example http_sd_config — #76
• Publish Docker images — #75
• Add last_successful_response metric — #78
• Add datacenter_info metric — #80, #81

Full Changelog: v6.1.0...v7.0.0-beta.1

v7.0.0-alpha.2

This is a pre-release which includes several significant changes.

• Breaking change — the -endpoint http://localhost:8080/metrics flag is replaced by -listen localhost:8080
• The /metrics route is no longer configurable
• Many new Compute@Edge metrics are now available
• Prometheus's generic HTTP service discovery is now supported
• The exporter now has a human-friendly-ish index page at the default / route
• "Official" Docker images are now available as ghcr.io/peterbourgon/fastly-exporter, please report any issues you may have

v6.1.0

This release upgrades the Prometheus dependency to avoid a benign but spurious and annoying WriteHeader error that can sometimes show up in the logs. Thanks @takanabe! You're a 🌟 There shouldn't be any breaking changes but I might have done messed up! It's the wild west out here on the OSS dude ranch.

v6.0.0

All the fun of v6.0.0-alpha.1 and v6.0.0-alpha.2, plus a new metric: pipe! Thanks @bridgetlane!

v6.0.0-alpha.2

This fixes a couple of bugs found in 6.0.0-alpha.1, specifically removing over-exposure (#58) and mis-mapping (#57) of edge_requests, and fixing misapplied histogram buckets (#56).

v6.0.0-alpha.1

This is an alpha or prerelease or something like that. It includes a fix for config files (#52) and a major overhaul of the way we parse fields from rt.fastly.com and map them to Prometheus metrics (#51). As part of this overhaul we include a lot more new fields, and also change the name of some metrics to be more Prometheus-standards-compliant. As a result this is a breaking change and that's why this is an alpha release. If current users could give it a try and report any major issues I would appreciate it.

v5.0.0

This release changes the flag parsing to use peterbourgon/ff, which gives us a couple of new features: most flags may now also be specified via env var and/or config file. It also renames the whitelist/blacklist options to allowlist/blocklist, which is a breaking change from v4.x.x. Finally, we fix a couple of small documentation typos. Enjoy!

v4.0.0

This release adds the ability to filter the Prometheus metrics that get exported (#40). It includes breaking changes to flag names for consistency.

v3.0.1

A fix for a small typo in the code that parses the -api-refresh-interval flag.

v3.0.0

First, for users with lots of services, especially if they receive lots of traffic, I've improved the cost of deserializing real-time stats by a factor of ten. This was the largest single contributor to CPU and memory pressure in the exporter.

Also, I've added some flags to help select services to be exported.

• name-exclude-regex: if set, ignore any service whose name matches this regex
• name-include-regex: if set, only include services whose names match this regex
• shard: if set, only include services whose hashed IDs modulo m equal n-1

More information in the README.

I've also added a few flags to tweak the network interaction between the exporter and the Fastly APIs.

• api-refresh: how often to poll api.fastly.com for updated service metadata (default 1m)
• api-timeout: HTTP client timeout for api.fastly.com requests (5–60s) (default 15s)
• rt-timeout: HTTP client timeout for rt.fastly.com requests (45–120s) (default 45s)

Thanks to @neufeldtech and @keur for their assistance with this release.