You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
IncidentIoIcingaWebStyle selects the Icinga Web 2 front end that source_url
points at: icingadb for Icinga DB Web, or monitoring for the older
monitoring module. Previously only Icinga DB Web was addressable, so the link
on every alert was a 404 for estates that have not migrated. Defaults to icingadb, and is read through globals so that leaving it unset is not a
configuration error.
SECURITY.md and CONTRIBUTING.md.
Changed
The Icinga Director basket now carries only the two custom variable fields, incident_io and incident_io_metadata. It previously shipped its own Command, User and NotificationTemplate, duplicating objects the package
already installs to conf.d with versions that could not work: the command
had no env block, so nothing reached the handler, and no notification was
ever applied to a host or service. Director cannot set env on a command at
all, so those objects were never going to be Director's to own. Setting vars.incident_io on Director-managed objects is the one thing Director is
needed for, and that is what the fields do. The orphaned DataList is gone
too; the boolean field never referenced it.
Fixed
make basket failed on a clean checkout, redirecting into a dist/ directory
that is gitignored and never created.
docs/TROUBLESHOOTING.md told operators to run the handler as nagios.
Icinga 2 runs as icinga on both Debian and RHEL families.
The example payload in the README omitted servicegroups, which service
alerts do carry.
Repository URLs still pointed at incident-io/icinga2, including the Homepage field baked into every package.
CI now checks that the version the handler reports matches the VERSION
file. Only the file was checked against the tag, so a release could ship
packages built as 0.2.0 whose --version and User-Agent still said 0.1.0.