v3.2.0

New

• Allow setting log level and formatting of logs as JSON - see README.md for new parameters

Potential Breaking Changes

• The changes to the logging in this release may break monitoring systems watching the output of the logs. Everything is now formatted with the go slog package which uses keys and values with the new option of JSON formatting

v3.1.3

NEW

• Log a warning if ups.status is not set to be exported because it is a variable most people will need. Thanks for the idea in #52, @varac!

MISC

• Fix documentation links for available variables. Thanks for the PR #51, @tigattack!
• Add link to a community ansible role. Thanks for the heads up in #53, @sleepy-nols!

v3.1.1

FIX

• The newly added nutserverport argument was also trying to parse NUT_EXPORTER_SERVER env variable - this was fixed to be NUT_EXPORTER_SERVERPORT. Thanks for the bug report in #43, @LXGaming!

v3.1.0

New

• Add support for setting a port for the NUT server - see updated usage in README. Thanks for the idea in #39, @mpeter!

Misc

• Fix local Dockerfile build process to avoid crashes due to libc differences. Thanks for the PR #41, @mpeter!

v3.0.0

Breaking

• The following command line parameters are no longer available, but can stil be configured in the new web.config.file parameter (with many others):
  • web.tls.cert_file
  • web.tls.key_file
  • web.auth.username

New

• TLS and web setup is now using the Prometheus exporter-toolkit. This enables many new options and parameters to control the listening endpoint. Thanks, @SuperQ for the PR in #13!
• The exporter should now build for riskv64 architectures now. Thanks for the suggestion in #31, @clausecker! Note: releases are not yet provided for this architecture

Misc

• The sample dashboard now allows you to select the datasource. Thanks for #30, @paulfantom

v2.5.3

Fixed

• Document the NUT_EXPORTER_WEB_AUTH_PASSWORD environment variable in help message. Thanks, @v4rakh!

v2.5.2

Fixed

• To improve performance, the previous release cached handlers/collectors based on UPS name. Instead, it should cache by server/upsName. Thanks for the report in #28, @JonyEpsilon!

v2.5.1

New

• Add an error condition if the requested UPS name is not found in the NUT server on first scrape
• Debug logging will now log the names of valid UPSs found in the NUT server - search for "UPS name detection"

Fixed

• Reolve a problem introduced in the previous release if the same UPS is scraped concurrently - thanks for the report in #27, @tsull360!

Misc

• Efficiency: If no UPS name is provided in the scrape, but one is found when listing UPSs, save the name so future scrapes don't have to list UPSs again
• Cleanup: Make the code a bit more readable

v2.5.0

New

• Add a new flag to disable the meta-metric network_ups_tools_device_info - thanks for the suggestion in #26, @kondas!

Fixed

• Fix the broken promhttp_metric_handler_requests_total metric due to how the registry/prom handler was built. Thanks for the report in #25, @kondas!

Misc

• Correct a few go lint warnings
• Add an example Grafana dashboard and screenshot to the dashboard directory

v2.4.2

Fixed

• Sometimes the underlying NUT library provides the wrong type of the variable. Don't rely on this and instead always rely on Go types