Skip to content

Releases: jonnenauha/prometheus_varnish_exporter

1.6.1

08 Dec 16:46
Compare
Choose a tag to compare
  • Fix duplicate counter errors on VLC reloads (#70 @LorenzoPeri)
  • Match cache hit rate calculation in example dashboard to match Varnish internals (#73] @dhpowrhost)
  • Start building releases with go 1.17.4

1.6

18 Dec 22:38
Compare
Choose a tag to compare
1.6
  • Varnish 6.5.x support (#64 @johanneswuerbach)
  • Update github.com/prometheus/client_golang dependency to v1.9.0
  • Remove GOARCH=386 builds from releases
  • Start building releases with go 1.15.6

1.5.2

27 Oct 11:59
Compare
Choose a tag to compare
  • Fix metric names and missing labels for file cache metrics (#55 @thedustin)
  • Fix scraping for Varnish 3.x. Removes the main_n_ban grouping. Metrics will now have individual bans_<type> metrics instad of the grouped metric that had type as a label. (#51 @glennslaven)
    • If you previously updated to 1.5 your exports would have already been broken as the grouping tries to combine gauge and counter metrics, which is not allowed by Prometheus.
    • This is breaking change if you are using Varnish 3.x and use ban metrics in your dashboards, you'll need to update them to the new ones.
  • Clean exported backend name if beginning with reload_ (#56 @stromnet)

1.5.1

01 Aug 12:59
Compare
Choose a tag to compare
  • Fix incorrectly typing Varnish 4.0.x stat flag a metrics as gauges instead of counters. (#48 @glennslaven)
  • Fix -test mode to wait for full metrics scrape before continuing.

1.5

24 Jun 08:48
Compare
Choose a tag to compare
1.5
  • Deprecate -no-exit. Default behavior is now not to exit on scrape errors as it should be for a long running HTTP server.
    • This was design misstep. You will now get a deprecation warning if you pass -no-exit but the process behaves as before.
    • New explicit -exit-on-errors has been added for users who want the old default behavior back.
  • Correctly export gauge and counter types from varnishstat output type property.
  • Add go module support.
  • Use github.com/prometheus/client_golang v1.0.0
  • Start building releases with go 1.12.6

1.4.1

10 Jul 11:54
Compare
Choose a tag to compare
  • -docker-container-name to signal that varnishstat should be ran in a docker container with docker exec <container-name> .
  • Support Varnish 6.0.0 by testing the main logic works and metrics are exported.
  • Start building releases with go 1.10.3

1.4

08 Dec 14:47
Compare
Choose a tag to compare
1.4
  • Standard non Varnish prometheus metrics need to now be enabled with -with-go-metrics. Before they were included by default. Now dropped to export less clutter that majority of users will never need (@nipuntalukdar).
  • Fix varnish_backend_up with Varnish 4.0 and earlier versions.

1.3.4

22 Oct 21:07
Compare
Choose a tag to compare
  • New per backend metric varnish_backend_up with 1/0 value that reflects the latest health probe result. The Varnish bitmap uint64 varnish_backend_happy as a prometheus float metric was not that useful in detecting latest up/down per backend.
  • Ability to give custom path to varnishstat with -varnishstat-path (by @zstyblik)
  • Github releases now include Grafana dashboards archive. This includes all the dashboards posted by users in the repo, starting with my own.

1.3.3

07 Oct 16:01
Compare
Choose a tag to compare
  • New -no-exit mode that does not exit the process if varnish is not running at the time of startup.
  • Support Varnish 5.2 that removed type and ident properties from varnishstat JSON output. If ident is not present, it is now parsed from the metric name.
  • Add tests to run scrape on static json files.
  • Start building releases with go 1.9.1

1.3.2

03 Sep 09:00
Compare
Choose a tag to compare
  • Update readme to mention that exporter has been tested to work against Varnish 5.x releases.
  • Start building releases with go 1.9