v0.16.0

⚠️ Breaking changes

• Added missing _total suffix to adfs and net collector counters (#667)
• Renamed mssql collector instance label to mssql_instance to prevent conflicts with the default Prometheus instance label (#691)
• Removed _total suffix from windows_dns_memory_used_bytes metric (#706)

Enhancements

• Experimental: Added smtp collector for the IIS SMTP service (#657)
• Experimental: Added dfsr collector for the DFS Replication service (#671)
• container collector now correctly sets prefix for container_id label for containerd (#716, thanks @jsturtevant!)
• Experimental: Added HTTPS support (#693)
• Added cache collector for the Windows file system cache (#723)
• Added cpu_info collector, exposing per-socket breakdown of system processors (#726, thanks @higels!)

Bug fixes

• Fixed issue with mssql collector where collector would always fail after an initial failure (#671)

Changes

• Clarified supported versions of Windows (#622)
• Updated Prometheus client library to v1.8.0 (#669)
• Added end-to-end testing script for CI (#672)
• Added common Prometheus SECURITY file (#699)
• Documented required version for time collector (#689)
• Fixed metric name in docs for windows_net_packets_outbound_discarded_total (#717, thanks @jsturtevant!)

v0.15.0

⚠️ Breaking changes

windows_mssql_transactions_active_total has been removed, and is replaced by windows_mssql_transactions_active (see below)

Enhancements

• Specific allowed remote IPs can now be set on the firewall rule created by the installed. The default is still to allow any machine access. (#608, thanks @siku4!)
• TCP collector now uses perflib instead of WMI, which will lead to faster collection times (#615, thanks @mallenLF!)
• TCP collector now exposed both IPv6 statistics in addition to the IPv4 statistics previously exposed. There is a new label af (address family) which lets you separate the two (#615, thanks @mallenLF!)
• The windows_exporter.exe file now has version information set (#563)
• Add a time collector with metrics about the Windows Time Service (#639)
• Add a collect[] url parameter if you need to customize what collectors to use per-scrape (#640, thanks @fischerman!)
• Add a --collectors.exchange.enabled flag to restrict which subcollectors are enabled on the exchange collector (#642, thanks @fischerman!)
• Add support for configuration files as a complement to command line flags (#638)

Bug fixes

• Fix a panic in the hyperv collector when WMI data is in the wrong format (#632)
• windows_mssql_transactions_active_total was erronously a counter, and has been replaced by a gauge windows_mssql_transactions_active (#636)

Changes

• The WMI library has been upgraded, which will hopefully fix rare memory leaks on some systems (#631, thanks @dcrn)
• Build exporter with Go 1.15 (#630, thanks @snoyberg!)
• Change the ordering of operations during upgrades, so that the old version is shut down before installing the new one (#597)
• Add a dependency on the WMI Performance Adapter for service start. This attempts to fix the exporter not starting at boot for some systems, please let us know if you still see the exporter failing to start (#645)

v0.14.0

Enhancements

• Add run_as label on windows_service_info (#527, thanks @paologallinaharbur!)
• Expose exporter version information on /version (#555, thanks @zlepper!)
• Convert MSSQL collector to use perflib rather than WMI (#592, thanks @mallenLF!)
• Add FSRM collector (#437, thanks @nwies!)

Bug fixes

• DHCP collector failure due to help mismatch (#533, thanks @breed808!)

Docs

Thanks to @JDA88, @zlepper, @clain23 and @eloekset for fixes and updates on the documentation since the last release!

v0.13.0

Breaking changes

Project renamed and moved

The exporter has been renamed to windows_exporter, and moved to the prometheus-community organization (see #519).

This includes changing the metrics' name prefixes from wmi_ to windows_, meaning that all queries need to be changed to match the new names.
For example, windows_cpu_time_total replaces wmi_cpu_time_total.

There are no functional changes in this release.

v0.12.0

Deprecation note

This is the last planned release with the wmi_exporter name, future releases will be named windows_exporter. See #499 and #519 for context.
This also means that the next release will have a breaking change, the metrics will change names from wmi_... to windows_.... You will need to adjust your dashboards and alerts. See this comment for a suggestion if you want to rename metrics already now.

Enhancements

• New collector: terminal_services (#491, thanks @asiyani!)
• New collector: remotefx (#491, thanks @asiyani!)
• New collector: dhcp (#493, thanks @breed808!)
• New collector: exchange (#508, thanks @rmyhren and @eikaas!)
• Add wmi_service_info metric with display_name and pid labels (#516, thanks @gsanchezgavier!)

Bug fixes

• Exporter now gives an error message instead of crashing if a non-existing collector is specified in --collectors.enabled (#529)
• Collector generator now works with Powershell 7 (#523, thanks @basroovers)

v0.11.1

Bug fixes

• ADFS collector was not registering correctly since 0.10.0, leading to failed scrapes (#506)

v0.11.0

Breaking changes

⚠️ This release has breaking changes! Make sure to read and understand the below before upgrading!

• We no longer redirect all paths to /metrics (#473). This means that if you configured Prometheus to scrape eg /not-metrics but did not configure wmi_exporter to expose the metrics on this port (using the --telemetry.path flag), your scrapes will now fail with a http 404. Make sure your Prometheus job configuration and the exporter configuration matches!
• The --collector.process.processes-where flag has been removed, in favour of --collector.process.whitelist and --collector.process.blacklist (see more below)

Enhancements

• The process collector has been rewritten to use perflib (#489), yielding significant performance improvements. Since it no longer uses WMI, the filtering mechanism can no longer use WQL. It now instead uses a white- and blacklist with regular expressions, like several other collectors.
• We now limit how many concurrent scrapes are allowed (#474). The exporter will only accept a limited number (by default 5) requests to be processed simultaneously. A flag --max-requests can be used to tune this. Typically, running several simultaneous scrapes tends to make the WMI subsystem slow.

v0.10.2

Bug fixes

• The version information is now properly written into the binary (#480)

v0.10.1

Bug fixes

• Fix cs collector crashing on domain joined machines (#476, thanks @sll552!)

v0.10.0

Enhancements

• Add wmi_os_info metric with information about Windows product name and version (#413, thanks @der-eismann!)
• Add wmi_cs_hostname metric with hostname, domain and fqdn info (#470, thanks @sll552!)
• Only query the perflib objects we need, improving performance and resolving some very slow queries on Exchange servers (#461)

Bug fixes

• Fix mssql collector ratio metrics (#463, thanks @secustor!)