Changelog¶
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
Unreleased¶
Highlights: A new check, rpm-database, warns when the RPM database is broken and the host silently stops receiving updates.
Added¶
Monitoring Plugins:
- freeipa-logfile: Warns when the daily ipa-healthcheck run finds a problem on a FreeIPA server, or no longer runs.
- icinga-director-basket-drift: Warns when Linuxfabrik objects in the Icinga Director differ from the basket of the installed plugins.
- icinga-director-overrides: Lists the service variables overridden per host in the Icinga Director, and overrides without effect.
- icinga-director-template-overrides: Lists where your own templates in the Icinga Director override the Linuxfabrik templates.
- rpm-database: Warns when the RPM database is broken, which stops dnf, yum and zypper.
Icinga Director:
- Icinga Director Service Set: Checks the Director daemon, its daily housekeeping and the Director checks above, on hosts tagged
icinga-director.
Changed¶
Monitoring Plugins:
- Log checks list what they alert on, repeated lines grouped and counted, and keep the first line short (logfile and the
*-logfilechecks for Apache, MySQL, PHP-FPM, Postfix and sshd). - about-me:
--tagsproposes the newdata-partitiontag on hosts with a partition mounted at/data, and the newicinga-directortag. - apache-httpd-logfile: Denied requests and path scans Apache rejects by itself no longer alert by default, other failed requests only from 10 on.
- lynis-logfile: Reports how many tests the lynis profile skips.
- mysql-connections: Speaks of failed connection attempts instead of aborted connections, and points to the denied logins.
- mysql-index-health: Redundant indexes no longer alert by default.
- mysql-innodb-buffer-pool-size: No longer alerts just because the data is larger than the buffer pool, and warns when sessions found no free page in it.
- mysql-memory: Alerts on the memory the database server really uses, the calculated worst case and other processes no longer alert.
- mysql-perf-metrics:
innodb_snapshot_isolationno longer alerts, and--storage-type=autono longer judges virtual disks. - mysql-temp-tables: No longer alerts by default, and no longer advises raising
tmp_table_size. - mysql-tls: A server on loopback only no longer alerts about
require_secure_transportand the certificate. - mysql-user-security: A legacy authentication plugin no longer alerts on MariaDB.
- nextcloud-stats: The output says "enabled apps", since disabled apps are not part of the count.
- php-fpm-logfile: An idle worker PHP-FPM kills while the pool shrinks no longer alerts, and a worker the OOM killer took is named as such.
- php-status: OPcache memory usage counts wasted memory, so a cache filling up with recompiled scripts alerts before it restarts. The restart warning sits next to the counter that triggered it.
- psi-cpu, psi-io, psi-irq, psi-memory: Alert on the five minute average instead of the last minute, so a job running for a few minutes no longer alerts.
- psi-io: Higher default thresholds, a single busy reader no longer alerts.
- rpm-updates: Counts Rocky Linux
securityrepo hot-fixes as security updates. - scanrootkit: Reports indicators instead of rootkits, generic paths and files any user can create only warn, detects 3 more implants.
- systemd-unit: Says why a failed unit failed, and shows its last log lines.
- systemd-units-failed: Says why each unit failed, and shows its last log lines when the check may read the journal.
Icinga Director:
- deb-updates, rpm-updates: Check for updates every 8 hours instead of once a day, and alert on a pending security update only after 30 hours.
- FreeIPA Server Service Set: Also checks the result of ipa-healthcheck and its timer.
- MariaDB and MySQL Metrics Service Sets: Also check long-running queries, which moved there from the Schemas Service Sets.
- needs-restarting: Alerts on a pending reboot or service restart only after 24 hours instead of 4 hours.
- OS - * Basic Service Sets:
Systemd Units Failedno longer alerts on a failedipa-healthcheck.service.
Grafana:
- re-import all dashboards and re-deploy the
icingaweb2-module-grafanaassets
Removed¶
Monitoring Plugins:
- mysql-memory: No longer counts the Galera GCache, including its perfdata.
- mysql-system: No longer checks
sunrpc.tcp_slot_table_entries, including its perfdata and Grafana panel.
Icinga Director:
- system-update Service Set: No longer checks
notify-and-schedule, which LFOps disables by default.
Fixed¶
Build, CI/CD:
- MSI: installing it over an earlier MSI updates the plugins; until now, the update was skipped without an error, and the host kept the plugins of its first MSI install
Monitoring Plugins:
- about-me: Rocket.Chat running as a rootless Podman container under
/opt/rocketchatgets its tag again, hosts with the LFOps system update are detected withoutnotify-and-schedule, and awikidirectory without MediaWiki no longer gets themediawikitag. - apache-httpd-logfile, mysql-logfile, php-fpm-logfile, postfix-logfile, sshd-logfile: An empty log right after a rotation no longer hides what the rotated log holds.
- dmesg: Ignores
SMBus base address uninitializedon Intel hosts too. - graylog-version: No longer times out right after a reboot on RHEL 8.
- mysql-index-health: Never suggests dropping a primary key as a redundant index.
- mysql-logfile: A MariaDB restart counts as one shutdown, and denied logins no longer come with "No errors or warnings found".
- mysql-memory: No more false alarms from a large
max_allowed_packetorjoin_buffer_sizeon MariaDB 11, and the Performance Schema memory is counted. - mysql-table-definition-cache: No longer counts the
information_schematables, which never take a slot in the cache. - needs-restarting: The first run after a reboot no longer breaks the RPM database on RHEL 8, and a broken RPM database returns UNKNOWN instead of a pending reboot.
- nextcloud-status: An HTTP 404 names the wrong path in
--url, and a web server error page is no longer quoted with its CSS. - rpm-updates: Lists packages with a dot in their name, such as
python3.13, under their own name and version.
v8.1.0 - 2026-10-27¶
Highlights: Mainly a bug-fix release: every plugin was tested again against all available Linux and Windows versions. New are aide-logfile for the AIDE file integrity check, lynis-logfile for the daily lynis audit, ntp-offset for the offset of the system clock and windows-version for the end of life of Windows. On RHEL, Debian and Ubuntu, the version checks follow the distribution's life cycle for the software it ships. Re-import all Grafana dashboards and the Director basket; hosts on Windows 10 or later and Server 2016 or later move to the per-version Basic Service Sets by their tag. The RPM and DEB packages move to the Linuxfabrik repository: install linuxfabrik-release once per host, or deploy it with the repo_linuxfabrik role of LFOps; the former repository at repo.linuxfabrik.ch/monitoring-plugins/ keeps receiving updates until 2026-12-31.
Added¶
Monitoring Plugins:
- aide-logfile: reports of the AIDE file integrity check
- lynis-logfile: the daily lynis audit
- ntp-offset: offset of the system clock, measured against an NTP server
- windows-version: end of life of Windows and Windows Server
Icinga Director:
- Fangfrisch Service Set
- InfluxDB Dump Service Set
- MongoDB Dump Service Set
- one Basic Service Set per Windows version, from Windows 10 to Server 2025
- PostgreSQL Dump Service Set
- system-update Security Lane Service Set
Assets:
- windows-jea: runs
procs,scheduled-taskandupdateswithout LocalSystem
Changed¶
Monitoring Plugins:
--timeoutdefaults to 3 seconds for a network connect or an API request, on 33 checks- end of life follows the distribution's life cycle, on most *-version checks
- spare, standby and offline components are evaluated, on all redfish-* checks
- the monitoring user needs its privileges on
*.*, on all mysql-* checks - the version perfdata changes its scale, on all *-version checks
- about-me: proposes the tags of the new Service Sets
- apache-httpd-logfile: more detail, far fewer false alarms
- cpu-vulnerabilities: fewer false alarms in virtual machines
- crypto-policy: compares only the base policy by default, and named subpolicies in any order
- deb-updates, rpm-updates: a timeout of apt or yum warns and names the command
- dns: the thresholds take Nagios ranges, honours
--timeoutalso on Windows, names the lookup time - gitlab-version: CRIT on a pending security update
- journald-usage: the thresholds take Nagios ranges, a plain number alerts only above it, UNKNOWN when journalctl fails
- memory-paging: swap-out no longer alerts by default
- mysql-connections, mysql-joins, mysql-memory, mysql-table-cache, mysql-temp-tables: fewer false alarms after a restart or on shared hosts
- mysql-health: fewer false alarms after a restart or on shared hosts, MariaDB no longer drops to 0/30 Security points
- mysql-index-health: unused indexes alert only with
--warning-unused - nextcloud-enterprise: shows how old the subscription data is
- ntp-chronyd: alerts on the real clock offset and on a silent NTP server
- ntp-ntpd: alerts on the real clock offset and on a silent NTP server,
--stratumapplies to this host - ntp-systemd-timesyncd: alerts on a silent NTP server,
--stratumapplies to this host - ntp-w32tm:
--warningand--criticalcheck the clock offset, renamed or dropped metrics, works on German Windows - php-fpm-logfile: reports fatal errors and hung workers, fewer false alarms
- php-fpm-ping: an unreachable ping page follows
--severity, perfdataresponse_timereplacesping - redfish-memory: fewer false alarms, detects defects on Fujitsu iRMC
- redfish-sensors: CRIT only above the fatal threshold, fewer false alarms, renamed power supply metrics
- restic-snapshots: warns when the repository has no snapshot or the filters match none
- scheduled-task: alerts on a failed last run, works on non-English Windows
- swap-usage: the thresholds take Nagios ranges, a plain number alerts only above it
- systemd-timedate-status: warns on an unsynchronized clock
- wordpress-*: find the installation without
--path
Icinga Director:
- Active Directory Domain Service Set: also checks Netlogon
- AIDE Service Set: uses
aide-logfile - Basic Service Sets: also check the clock offset with
ntp-offset - FreeIPA Server Service Set: also checks its certificates
- Lynis Service Set: uses
lynis-logfile - OS - Debian * Basic Service Sets: combine with the networking Service Set
- OS - Windows Basic Service Set: renamed to "Windows 8/2012 or below"
- Postfix MTA Service Sets:
Postfix Versionalerts on end of life - system-update Service Set: also watches
update-and-reboot.timer - WordPress Service Set: finds the installation itself
Grafana:
- re-import all dashboards and re-deploy the
icingaweb2-module-grafanaassets
Build, CI/CD:
- RPM, DEB: the packages come from the Linuxfabrik repository, which the
linuxfabrik-releasepackage sets up; repo.linuxfabrik.ch/monitoring-plugins/ receives the updates until 2026-12-31 and is switched off then
Removed¶
Monitoring Plugins:
- redfish-storage: perfdata
<drive>_power_on_hours
Fixed¶
Monitoring Plugins:
- a hanging command returns WARN after
--timeout, on many plugins - clear messages, no Python tracebacks or raw OS errors, on many plugins
- compressed logs are no longer reported again on every run, on logfile, apache-httpd-logfile, mysql-logfile, php-fpm-logfile, postfix-logfile and sshd-logfile
- HTTPS works with root certificates Windows has not cached yet, on all plugins
- non-ASCII arguments are no longer garbled, on all Windows plugins
- the acknowledgement is read from the right service, on all plugins with
--icinga-callback - the message names a threshold that is no Nagios range, on all plugins with ranges
- apache-httpd-logfile, mysql-logfile, php-fpm-logfile, postfix-logfile, sshd-logfile: no journal warning without systemd (#1544)
- borgbackup: a running backup is no longer UNKNOWN
- by-ssh:
--warning-regexand--critical-regexignore case, as documented - by-winrm: works with JEA endpoints and with pywinrm only, regexes ignore case
- cert:
--source=fileworks on RHEL 8 - cpu-usage: no alert on idle CPUs waiting for I/O
- csv-values, json-values: a clear UNKNOWN without
--filenameor--url, or on invalid JSON - deb-lastactivity, rpm-lastactivity: the notification for a failing package database names the reason
- dhcp-scope-usage: works again, also locally on Windows
- disk-io: learns the bandwidth for a week before it warns
- disk-usage, nfs-exports, nfs-mounts, path-rw-test: return on time despite a hung NFS mount
- dmesg: fewer false alarms on Debian 13 and Ubuntu 26.04 VMs
- docker-image, podman-image: no false CRIT for images dated 1970
- fedora-version, rhel-version: no false end of life on other distributions
- feed: respects the time zone of an entry, acknowledging an entry without an ID no longer silences the next one
- file-age, file-count, file-growth, file-size: more robust on SMB shares and with files that come and go
- file-ownership: no false warning for unbound on Fedora 44
- fs-inodes: checks every logical volume
- fs-ro: no missed read-only mount
- grafana-version: reads the version more reliably
- grassfish-screens: two Grassfish servers checked from the same host no longer share their screen data
- graylog-version: a removed package that left its configuration behind on Debian no longer counts as installed
- haproxy-status, starface-*, ups-nut: no garbled non-ASCII text
- huawei-pacific-quota: warns on an incomplete quota listing
- infomaniak-swiss-backup-devices: works behind a mandatory proxy
- journald-query: UNKNOWN without journal rights or when journalctl fails, reads binary messages
- mediawiki-version: reads MediaWiki before 1.31.7
- moodle-version: finds Moodle 5.1 and newer
- network-errors: one state file for all interfaces, also cleans up after interfaces that went away
- nextcloud-version: no longer hangs when
occdoes not answer - openjdk-redhat-version: no wrong version with quoted Java options
- podman-stats: no false CRIT for rootless containers on cgroups v1
- postfix-logfile: reads the journal on Debian and Ubuntu
- qts-cpu-usage: keeps a separate history per appliance
- qts-disk-smart: works with one disk and on QTS 4.2, keeps CRIT on overheating
- qts-temperatures: works on models without CPU temperature
- qts-version: reports updates again (#1547)
- redfish-logservices: reads the SEL of HPE, names an unreadable log
- redfish-storage: no UNKNOWN on controllers that send empty drive or volume lists
- redis-version, valkey-version: no longer confuse Redis and Valkey
- sshd-logfile: detects MaxStartups rate limiting again
- systemd-units-failed: no longer names an ignored unit, no longer waits on a slow journal
- tuned-profile: accepts post-loaded and System Roles profiles
- updates: warns on a disabled Windows Update service, no longer hangs (#695)
- users: no longer counts the per-user systemd service manager as an SSH session on RHEL 10
- wildfly-version: honours
--proxyand--timeoutfor the GitHub lookup - xca-cert: works without
--prefix
Notification Plugins:
- notify-host-mail, notify-service-mail: no error on a line break in the subject
Icinga Director:
- Active Directory Lightweight Directory Service Set: checks the AD LDS instances
- apache-httpd-security, deb-updates, redfish-storage, rpm-updates, snmp: re-import the basket
- by-ssh, by-winrm, dhcp-scope-usage: re-import the basket, so that a hanging command returns WARN
- ClamAV Service Set: works on Debian and Ubuntu
- FortiOS 6 Service Set:
FortiOS Sensorgets the host address - IPMI, kdump, Nextcloud, Scanrootkit and all OS Basic Service Sets: deb-updates, dmesg, ipmi-*, kdump, nextcloud-security-scan and scanrootkit run via sudo
- Mastodon and Rocket.Chat Service Sets: check the rootless containers
- Nextcloud Service Set: no warning for an enabled
nextcloud-app-update.timer - ntp-chronyd, ntp-ntpd, ntp-systemd-timesyncd, ntp-w32tm: re-import the basket
- PHP-FPM Service Set: no false alert on Debian and Ubuntu
- sap-open-concur-com: re-import the basket, so the times follow daylight saving time
- Windows service sets: a stopped service is no longer hidden
Build, CI/CD:
- RPM: no more
EOFError: marshal data too short(#1543)
Security¶
Monitoring Plugins:
- other local users can no longer plant state files on Windows, on all plugins
- acmesh-status, docker-service, docker-swarm: no longer reveal which files exist
- apache-httpd-logfile, mysql-logfile, php-fpm-logfile, postfix-logfile, sshd-logfile: a swapped file, directory or symlink reveals nothing outside
/var/log - apache-httpd-security, nginx-security:
--commandruns only root-owned system binaries - fail2ban:
--socketmust be root-owned - icinga-topflap-services: never shows the password file, ignores downtimes (#1564)
- logfile: cannot escape
/var/log, works on Windows again (#1587) - lynis:
--lynis-optioncannot point lynis at files or programs, lists every finding, refuses to run as root - mastodon-version:
--pathno longer follows symlinks out of the installation, no false CRIT when Podman fails - metabase-stats: no session token for another instance, works again
- mysql-logfile:
--defaults-filehas to belong to the account calling sudo - mysql-user-security: no longer prints a weak password it found
- nextcloud-*:
--pathrunsocconly from a root-controlled installation - nextcloud-security-scan: accepts only a well-formed scan ID from the scanner
- openvpn-client-list: confined to
/var/log, supports OpenVPN 2.7 and IPv6, a plain threshold alerts only above it - php-status: reads only root-owned pool configs,
--configcompares the whole value, finds every missing module - podman-*:
--userneeds a subordinate UID range - postgresql-version:
--pathruns only apsqlthat nobody but root can change - redis-status, valkey-status: the password stays out of the process list, a refused login is UNKNOWN, a recovered replica resync no longer warns until the next restart
- restic-*:
--repocan no longer run programs or write files as root, works with restic 0.18 - scanrootkit: no longer reveals which files exist, detects 23 more rootkits and those named by a pattern
- snmp: a device answer is never run as code, works with the default
--deviceand on Debian and Ubuntu - strongswan-connections:
--socketmust be root-owned (GHSA-cw8h-7h72-79v8) - systemd-unit:
--machinetakes only a container name
v8.0.0 - 2026-09-23¶
Highlights: A hung network filesystem no longer takes a check down with it. More than thirty new checks cover LVM, software RAID, multipath, NFS, KVM guests, pressure stall information, server logs and the hardening of Apache httpd and NGINX. Several Director parameters, host tags and Service Sets change, so read the Breaking Changes first. fail2ban and kdump close a local privilege escalation.
Breaking Changes¶
Monitoring Plugins:
- file-age, file-size: no longer run through sudo, grant the monitoring user read access where needed
- huawei-dorado-controller:
--warningand--criticalcover CPU only, set--warning-memand--critical-memfor memory - huawei-dorado-port: checks front-end and cluster ports only, add
--include-backendfor the others - podman-stats: block and network I/O are rates under new metric names, adjust your graphs (#1519)
- sudoers: split in two, a hand-deployed setup also needs
*-logging.sudoers(#1493) - whmcs-status:
--urlis required, and a missing health check no longer counts as healthy
Icinga Director:
- Deb Updates, RPM Updates: also alert on ordinary updates after a grace period, tick
Only Criticalfor the old behaviour - delete the leftover
tpl-service-certthat usescmd-check-url(#1474) - Huawei Dorado Service Set: a Director that imported the v7.0.0 basket needs five services renamed before the import
- Journald Query, Logfile, MySQL Logfile:
Ignore Pattern/Ignore RegexbecomeIgnore(regex), and MySQL Logfile'sServer Loghas to be re-entered - KVM Host Service Set: tag your hypervisors
libvirtdorvirtqemud - MySQL Database Metrics, Storage Engines, Table Indexes:
Ignore Schemas/Ignore TablesbecomeMatch/Ignore - the
rpm-updatestag and Service Set are gone, the check is part of the Basic Service Sets
Added¶
Monitoring Plugins:
- acmesh-status: expiring acme.sh certificates and stalled renewals
- apache-httpd-disclosure: what an Apache httpd server gives away about itself (#373)
- apache-httpd-logfile: problems in the Apache httpd error log
- apache-httpd-security: hardening of a local Apache httpd (#373)
- avelon-tickets: open alarm tickets in the Avelon Cloud (#770)
- conntrack: a filling netfilter connection tracking table
- cpu-vulnerabilities: unmitigated CPU vulnerabilities
- file-growth: a file growing or shrinking too fast (#48)
- fs-mounts: filesystems from
/etc/fstabthat are not mounted - huawei-dorado-quota: shares filling up their quota
- kvm-cpu-usage: CPU usage and steal time per virtual machine (#644)
- kvm-disk-io: disk I/O and storage latency per virtual machine (#644)
- kvm-memory-usage: memory usage per virtual machine (#644)
- kvm-network-io: network I/O and drops per virtual machine (#644)
- kvm-storage-pool: state and free space of libvirt storage pools (#644)
- kvm-volume: libvirt volumes and pool overcommitment (#644)
- lvm-snapshots: filling and invalidated LVM snapshots
- lvm-thin-pools: LVM thin pools running out of data or metadata
- lvm-volume-groups: missing physical volumes and free space of LVM volume groups
- lvm-volumes: incomplete, degraded or inactive LVM logical volumes
- md-raid: software RAID arrays losing redundancy
- memory-paging: swap paging activity
- metabase-version: Metabase end of life and new releases
- multipath: LUNs losing paths
- nfs-exports: configured NFS exports that are not served
- nfs-mounts: stale or unresponsive NFS mounts
- nginx-disclosure: what an NGINX server gives away about itself
- nginx-security: hardening of a local NGINX
- openstack-cinder-list: block storage volumes of a project
- openstack-quota: compute, block storage and network quotas of a project (#489)
- php-fpm-logfile: problems in the PHP-FPM error log
- postfix-logfile: problems in the Postfix mail log
- psi-cpu: work waiting for a CPU (#746)
- psi-io: work waiting for storage (#746)
- psi-irq: CPUs busy servicing interrupts (#746)
- psi-memory: work waiting for memory (#746)
- sshd-logfile: failed logins and problems in the OpenSSH server log
Icinga Director:
libvirtd Service SetLVM Service SetMD RAID Service SetMultipath Service SetNFS Client Service SetSensors Service Setsmartmontools Service Setvirtqemud Service Set
Changed¶
Monitoring Plugins:
- apache-httpd-status: more accurate worker usage, and works with
ExtendedStatus Off - cpu-usage: alerts on CPU steal, and correct percentages on hosts running virtual machines
- dmesg: fewer false alarms on physical servers and in virtual machines
- docker-*, podman-*: an unresponsive container engine warns after
--timeout, and--always-okcovers it - huawei-dorado-host, huawei-dorado-interface, huawei-dorado-lun: adjusted severities
- kvm-vm: reports crashed machines, and no longer needs root
- lynis: alerts when no host was audited
- mysql-database-metrics, mysql-storage-engines, mysql-table-indexes:
--ignore-schemas/--ignore-tablesare deprecated - mysql-innodb-buffer-pool-size, mysql-innodb-log-waits: no more redo log alerts on idle databases
- mysql-logfile: aborted connections and denied logins are counted per source over time instead of alerting on each
- nextcloud-enterprise: alerts on an expired subscription and on account limits (#647)
- nextcloud-stats: lists the five largest accounts, which is slow on large instances,
--top=0turns it off (#103) - openstack-nova-list: alerts on an ACTIVE instance that is not running, and fewer false CRITICALs
- openstack-swift-stat: alerts on container and account quotas
- scanrootkit: detects 13 more rootkits and implants
- wordpress-security-scan: warns when it cannot check for vulnerabilities
Icinga Director:
- the Needs Restarting Service Set covers Debian, tag those hosts
Grafana:
- re-deploy the
icingaweb2-module-grafanaassets: Icinga Web 2 shows all graphs of a check - re-import the dashboards of apache-httpd-status, cpu-usage, disk-io, huawei-dorado-hypermetropair, huawei-dorado-lun, Icinga overview, keycloak-memory-usage, kvm-vm, load, memory-usage, mysql-logfile, network-io, php-status, ping, procs and swap-usage
Removed¶
Monitoring Plugins:
- huawei-dorado-hypermetropair, huawei-dorado-lun: the per-object status code metrics
- swap-usage: the
sinandsoutmetrics, see memory-paging
Icinga Director:
- the Basic and Apache Service Sets for Debian 10, RHEL 7 and Ubuntu 16 to 20, including their host tags: retag those hosts
- the
File Size - /var/log/audit/audit.logservice in the RHEL and Fedora Basic Service Sets
Build, CI/CD:
- packages for Ubuntu 20.04
Fixed¶
Monitoring Plugins:
- about-me: detects KVM hosts with the modular libvirt daemons again
- all checks over HTTP: honour network ranges in
no_proxyand the Windows proxy exceptions - all plugins: a command stuck on vanished storage no longer runs past
--timeout - all
*-versionchecks: releases missing on endoflife.date no longer cause UNKNOWN or a crash - cert:
2wworks as a threshold, and proxy settings are honoured (#1474) - deb-updates:
--only-criticalno longer misses a fresh security update, and concurrent runs no longer mix results - disk-io: no false warning after a reboot (#677)
- disk-usage:
--fstypeworks on a host with a hung network filesystem - docker-service, docker-swarm: can reach the Docker daemon via sudo
- docker-stats:
--countholds for every container on busy hosts - file-ownership: a malformed
--filenameno longer crashes the check - fortios-network-io, fortios-sensor, jitsi-videobridge-status:
--always-okworks - gitlab-version: an unreachable version-check service follows
--unreachable-severityinstead of forcing UNKNOWN - haproxy-status: performance data no longer breaks on failed health checks or tracked servers
- huawei-dorado-*: fewer false alarms and missed faults, correct I/O sizes, and no timeouts on large arrays
- icinga-topflap-services, kubectl-get-pods: concurrent runs no longer mix results
- keycloak-memory-usage, keycloak-stats, keycloak-version: name the missing "manage-realm" role instead of crashing
- logfile: a log that is not valid UTF-8 no longer breaks the check
- needs-restarting: fewer false results on RHEL and Debian (#1522)
- nextcloud-status, spring-boot-actuator-health: honour proxy settings (#1474)
- php-status: no longer warns when
post_max_sizeis smaller thanupload_max_filesize - redfish-*: recover after the controller drops its sessions, and put far less load on it (#1372, #1507, lib#350)
- redfish-logservices: evaluates the System Event Log of Avigilon servers
- rpm-updates: shows the correct target version, and concurrent runs no longer mix results
- rpm-versionlock: finds all dnf 5 locks (#1462)
- users: counts disconnected sessions on Windows
- wildfly-non-xa-datasource-stats, wildfly-xa-datasource-stats:
--always-okalso covers a missing datasource - wordpress-checksums: no false alarm on localized installations, and proxy settings are honoured (#1474)
- xml: a missing
lxmlmodule reports UNKNOWN instead of a stack trace
Icinga Director:
- the Huawei Dorado Service Set runs all its checks again and copes with large arrays: re-import the basket
- the Postfix MTA Service Sets no longer abort
basket restore
Build, CI/CD:
- the SELinux policy loads on RHEL 10 again
Security¶
Monitoring Plugins:
- all checks over HTTP: passwords no longer leak via redirects or error messages, and oversized answers are refused (GHSA-pq9x-4pp3-p5r9)
- fail2ban:
--socketis confined to/run, closing a local root code execution - kdump:
--pathno longer discloses files outside the crash-dump directory (GHSA-q8c8-wxhc-3h4c)
v7.0.0 - 2026-08-14¶
Highlights: disk-io stops raising false CRITICALs on ZFS and Proxmox, and the Redfish checks no longer time out on large servers. Thirty-five new checks cover Docker, Podman, Huawei OceanStor, WordPress and package version locks. Counters become per-second rates and redfish-* needs --url, so re-import dashboards and review commands before updating.
Breaking Changes¶
Monitoring Plugins:
- counters are per-second rates with partly new metric names: re-import the dashboards of cpu-usage, disk-io, fs-xfs-stats, jitsi-videobridge-stats, network-io, nginx-status, nodebb-cache, nodebb-errors, procs, redis-status, starface-database-stats, valkey-status and wildfly-gc-status (#320)
- disk-io: no longer measures I/O wait (false CRITICALs on ZFS and Proxmox) and only warns, re-import the dashboard (#1371)
- docker-stats, podman-stats: special characters in per-container metric names become
_(web.1→web_1) - huawei-dorado-*: metric names and units changed, re-import the dashboards
- redfish-*:
--urlis mandatory (#1306)
Icinga Director:
- the Host and Service templates are pinned to the master zone: with a satellite tier, unset the zone on
tpl-host-genericandtpl-service-generic(#721)
Added¶
Monitoring Plugins:
- apache-tomcat-version: Apache Tomcat end of life and updates (#126)
- deb-versionlock: packages APT holds back
- docker-container, podman-container: unhealthy, restarting or unexpected-state containers
- docker-image, podman-image: outdated images
- docker-service: Docker Swarm services running too few tasks
- docker-swarm: swarm membership, down nodes and manager quorum
- huawei-dorado-alarm: current alarms
- huawei-dorado-expboard: faulty expansion boards
- huawei-dorado-lun: faulty and filling LUNs
- huawei-dorado-port: faulty or slow front-end ports
- huawei-dorado-sfp: faulty optical modules and light levels
- huawei-dorado-storagepool: faulty and filling storage pools
- huawei-pacific-alarm: current alarms
- huawei-pacific-disk: faulty and worn-out disks
- huawei-pacific-fan: faulty fans
- huawei-pacific-namespace: unreachable or read-only namespaces
- huawei-pacific-node: faulty cluster nodes and expired warranties
- huawei-pacific-power: faulty power supplies
- huawei-pacific-quota: shares filling up their quota
- huawei-pacific-replicationpair: replication pairs that stopped mirroring
- huawei-pacific-service: stopped service processes
- huawei-pacific-storagepool: faulty and filling storage pools
- huawei-pacific-system: cluster capacity usage
- icingaweb2-module-updates: outdated Icinga Web 2 modules (#124)
- kdump: whether a kernel panic can be captured, and leftover crash dumps
- librenms-validate: problems LibreNMS reports about itself (#366)
- network-errors: interface receive and transmit errors (#707)
- nextcloud-app-updates: pending Nextcloud app updates (#62)
- nextcloud-status: pending database upgrades and maintenance mode (#329)
- rpm-versionlock: packages the RPM package manager holds back
- wildfly-version: outdated WildFly (#123)
- wordpress-checksums: modified WordPress core and plugin files
- wordpress-security-scan: known vulnerabilities and exposed credentials on a WordPress site
Icinga Director:
Icinga Web 2 Service SetLynis Service Set(tag only the host that runs the subnet audits)OpenJDK Service Set- host tag
metabase(no Service Set, apply the service template yourself)
Assets:
- bash completion for the plugin options
Changed¶
Monitoring Plugins:
- all plugins: output shows
<,>and&verbatim instead of escaped - cpu-usage: no longer alerts on iowait
- docker-info: reports every warning the daemon raises
- huawei-dorado-*: faulty components are CRITICAL instead of WARNING, and large arrays are fully reported
- mysql-innodb-log-waits: alerts only on real log waits
- php-status: warns when
post_max_sizeis not larger thanupload_max_filesize(#516) - podman-stats: CPU usage is measured since the last run instead of since container start
- rhel-version: points to fedora-version on Fedora
- scanrootkit: detects VoidLink and RingReaper
- uptimerobot: checks UptimeRobot's own status page by default, other pages need
--url
Icinga Director:
- huawei-dorado-disk, huawei-dorado-host, huawei-dorado-hypermetropair: hide items within their thresholds, re-import the basket
- the WordPress Service Set and host tag follow WordPress' spelling, re-tag your hosts
Fixed¶
Monitoring Plugins:
- about-me: detects WordPress in the document root
- borgbackup, file-ownership, getent, nextcloud-enterprise, rpm-lastactivity, scheduled-task: work again
- cert: a subnet scan finishes within the timeout and needs far less memory
- csv-values, json-values: non-UTF-8 input no longer crashes the check (lib#256)
- deb-lastactivity: no stack trace on a host without APT packages
- disk-smart: reads RAID and USB drives again,
--ignoreworks, and a failing drive stays CRITICAL (#1388) - disk-usage: performance data carries the thresholds again, and
(?-i:...)patterns match (#1310) - docker-stats: a container without statistics no longer causes UNKNOWN
- fs-inodes: an unreadable mount point no longer aborts the check (#1387)
- journald-query: relative
--sincevalues work again (#1264) - librenms-health: sensors past their limits alert
- logfile: detects a logfile that gets rewritten from the start (#1330)
- lynis: works where lynis lives outside
/usr/share(#1262) - mysql-replica-status: works on MySQL 8.4
- mysql-user-security: the suggested
ALTER USERworks on MariaDB 11.6 and newer - ping: corrupted packets are counted correctly
- podman-info: no crash without unqualified search registries
- redfish-*: no timeouts on large servers (#1372)
- sensors-fans, sensors-temperatures: identical sensors no longer overwrite each other's performance data
- snmp: net-snmp warnings no longer abort the check, and string-indexed OIDs work
- statusiq: no more flapping to UNKNOWN
- strongswan-connections: no false alarms or crashes on rekeying, shared, connecting or 3DES connections (#806)
- systemd-unit: the Ubuntu Service Sets check
ssh.service(#1373)
Grafana:
- ping: times are shown in milliseconds, re-import the dashboard
Assets:
- SELinux policy: loads on RHEL 10
Security¶
Monitoring Plugins:
- all checks over HTTP: API keys and session tokens no longer follow a redirect to another host (GHSA-4jc5-g844-4x33)
- all plugins:
--testcan no longer read root-owned files via sudo (GHSA-rh9c-rqvg-f7pr) - keycloak-memory-usage, keycloak-stats, keycloak-version: admin credentials no longer leak to another host (GHSA-88fj-95f7-w68m)
- logfile: closed a local privilege escalation (only with
fs.protected_symlinks=0) (GHSA-w2gg-hx6w-24w3) - logfile, mysql-logfile, openvpn-client-list: log files are confined to
/var/log, bind-mount logs stored elsewhere (GHSA-f54c-p5vg-mr5c) - redfish-*: a malicious controller can no longer redirect a check to another host (GHSA-96fx-pqc3-28xv)
- virustotal-scan-url: the API key can no longer leak to another host
Notification Plugins:
- notify-host-mail, notify-service-mail: a monitored service can no longer inject markup into the email
v6.0.0 - 2026-06-14¶
Highlights: A local privilege escalation through crafted plugin arguments is closed. The Redfish checks are renamed, so update your commands, and five new ones join them.
Breaking Changes¶
Monitoring Plugins:
- redfish-*: renamed to match their API endpoints (
redfish-drives→redfish-storage,redfish-sel→redfish-logservices,redfish-sensor→redfish-sensors,redfish-system→redfish-systems), update your commands
Added¶
Monitoring Plugins:
- lynis: security hardening of the hosts in a subnet, over SSH
- redfish-ethernetinterfaces: Ethernet interface health
- redfish-firmwareinventory: firmware versions and health
- redfish-managers: management controller health (iLO, iDRAC)
- redfish-memory: memory module health
- redfish-processors: processor health
Changed¶
Monitoring Plugins:
- by-ssh:
--shellis ignored, pipes, globs and variables always work - cert: scans whole subnets and checks the full certificate chain
- ipmi-sensor: performance data is grouped by sensor type, which resets the graph history once (#22)
- nextcloud-security-scan: fresh rating right after a Nextcloud update (#118)
- php-status: OPcache warns at 95% and on cache thrashing
- redfish-*: no longer flood the controller's session table and audit log, and retry flaky requests
- swap-usage: a host without swap is OK instead of UNKNOWN (#1142)
Fixed¶
Monitoring Plugins:
- about-me: no crash while detecting installed software
- apache-httpd-version: works again (PR #1224, thanks to Salman Mohammadi)
- by-ssh: a failed connection no longer echoes the
--passwordvalue - redfish-sensors: no false warnings from placeholder ranges (#1211)
- several plugins running system commands: a harmless warning on stderr no longer causes UNKNOWN
- Windows: no blank lines between output lines, and no garbled umlauts (#681)
Icinga Director:
- checks deploy correctly in distributed setups, the templates no longer pin them to the master zone (#721)
Security¶
Monitoring Plugins:
- all plugins: crafted arguments can no longer execute arbitrary commands, most seriously via sudo (GHSA-798h-hpph-m24j)
v5.2.0 - 2026-06-02¶
Highlights: Trend-data caches move out of the shared /tmp, closing a local symlink attack on checks running as root.
Security¶
Monitoring Plugins:
- plugins caching trend data no longer use
/tmp, closing a local symlink attack (GHSA-r35r-fpx2-jgr4, thanks to OoYo0uto)
v5.1.0 - 2026-05-30¶
Highlights: A local privilege escalation through the Debian apt-get sudoers rule is closed. The mysql-* checks raise fewer false alarms, and about-me detects more platforms.
Changed¶
Monitoring Plugins:
- about-me:
--tagsdetects more software and platforms, and all user-installed packages are listed - fail2ban: thresholds accept Nagios ranges, and a banned jail no longer mislabels the others (#140)
Fixed¶
Monitoring Plugins:
- all plugins: no longer abort on RHEL 8's default Python 3.6
- mysql-*: no more "Illegal mix of collations" (#1139)
- mysql-innodb-buffer-pool-size: works on MySQL 9.3 and newer
- mysql-perf-metrics: fewer false alarms on MySQL 9.0 to 9.2 and on network storage
- mysql-table-definition-cache: recommends a valid value
- snmp: a malformed threshold in a device CSV reports UNKNOWN instead of being ignored (#768)
Security¶
Assets:
- Debian sudoers: the
apt-getrule no longer grants a root shell (GHSA-8w6w-23mq-h8rg, thanks to OoYo0uto)
v5.0.0 - 2026-05-15¶
Highlights: The mysql-* family is reworked: counters become rates and perfdata labels change, so re-import every MySQL dashboard. Six checks are new, two are gone.
Added¶
Monitoring Plugins:
- cert: expiring X.509 certificates on TLS endpoints and in local files
- mysql-health: a single 0-100 health score for MySQL/MariaDB
- mysql-index-health: unused and redundant indexes
- mysql-long-queries: long-running queries
- mysql-tls: TLS setup and certificate expiry of MySQL/MariaDB
- ups-nut: UPS managed by Network UPS Tools
Changed¶
Monitoring Plugins:
- mysql-*: counters become rates and perfdata labels change, re-import the dashboards; thresholds use Nagios ranges (
>=Nbecomes>N) - mysql-aria, mysql-binlog-cache, mysql-innodb-log-waits: a disabled engine or
log_bin = OFFis OK instead of UNKNOWN - mysql-innodb-buffer-pool-size: checks
innodb_redo_log_capacityandinnodb_file_per_table - mysql-logfile: reads the error log from Performance Schema where available, and reads container logs again
- mysql-memory: counts the Galera GCache, alerts at 85% and 95%, and accounts for
max_tmp_table_sizecorrectly - mysql-perf-metrics: a deprecated variable only warns when set explicitly
- mysql-replica-status: needs fewer privileges on MariaDB, and no longer reports lag on every server
- mysql-storage-engines: the AUTO_INCREMENT check respects each column's type
- mysql-system: warns on
fs.nr_open < 1M, and perfdata is renamed - mysql-table-indexes: flags InnoDB tables without a primary key
- mysql-user-security: flags legacy authentication plugins and default passwords
Icinga Director:
- mysql-binlog-cache moves to the MySQL Service Set, hosts with only
mysql-replicationalso needmysql
Fixed¶
Monitoring Plugins:
- docker-stats, podman-stats: per-container CPU and memory perfdata is back (#1104)
- mysql-database-metrics: no longer misjudges the
perconaschema or index sizes - mysql-slow-queries: a ratio of 5.x% alerts again
- mysql-temp-tables: no crash on idle servers
- mysql-thread-cache: correct perfdata unit
- mysql-traffic: no "100% writes" on idle servers
- veeam-status: works with Veeam Enterprise Manager v13 (#1001)
Grafana:
- dashboards import into Grafana 12 again
Removed¶
Monitoring Plugins:
- hin-status: the HIN status page no longer exists
- mysql-innodb-buffer-pool-instances: obsolete on current MariaDB and MySQL
v4.1.0 - 2026-05-08¶
Changed¶
Monitoring Plugins:
- sap-open-concur-com: defaults to the
eu2datacenter, and slow responses no longer cause UNKNOWN
Removed¶
Icinga Director:
- 13 single-plugin Service Sets, apply their templates via Apply rules
- the
tarifpool-v2host tag
Fixed¶
Monitoring Plugins:
- network-port-tcp: no longer crashes
- php-fpm-status: no false CRIT on dynamic and ondemand pools
v4.0.0 - 2026-05-07¶
Added¶
Icinga Director:
Needs Restarting Service SetOS - RHEL 10 Basic Service SetPostfix MTA Service Set (Multi-Instance)(#535)
Changed¶
Monitoring Plugins:
- dmesg:
--ignoretakes regexes and replaces the defaults,--severityis ignored and alerts are always CRIT
Icinga Director:
- the Basic Service Sets no longer check
rsyslog.service, tag rsyslog hostsrsyslog
Removed¶
Icinga Director:
- the
OS - Debian 8 Basic Service Set
Fixed¶
Monitoring Plugins:
- librenms-alerts:
WORSE,BETTERandCHANGEDalerts are no longer reported OK
v3.0.0 - 2026-05-05¶
Breaking Changes¶
Monitoring Plugins:
- plugins with repeatable parameters: your values replace the defaults instead of extending them (#540)
- haproxy-status:
--username/--passwordare replaced by credentials in--url - mailq: thresholds take a duration instead of a count (#781)
- php-fpm-status: all perfdata labels are renamed and prefixed
<pool>_, update your queries - procs:
--argument,--commandand--usernametake regexes, use^foo$for an exact match - redfish-sensor:
--insecureis the default, pass--insecure=falsefor trusted certificates
Tools:
check2basketis nowbuild-basket,remove-uuidsis nowbasket-remove-uuids
Added¶
Monitoring Plugins:
- by-winrm: runs commands on Windows hosts via WinRM
- nextcloud-enterprise: Nextcloud Enterprise subscription
- podman-info: system-wide Podman information (#1023)
- podman-stats: CPU and memory of Podman containers (#1023)
- redfish-system: overall system health (#652)
Icinga Director:
- Debian 13 Service Set
- Ubuntu 26 Basic Service Set
Build, CI/CD:
- packages for SLE 15, SLE 16 and Ubuntu 26.04
Changed¶
Monitoring Plugins:
- all plugins: unknown arguments are ignored instead of erroring
- atlassian-statuspage: the
impactperfdata is renamed tocnt_warnandcnt_crit - disk-io: also alerts on iowait
- file-count: much faster on large directories
- file-ownership: checks more CIS-relevant files by default, which may raise new alerts
- gitlab-version: warns on security updates (#688)
- nextcloud-version:
occno longer has to be executable - php-status: defaults to
http://localhost/monitoring.php - scanrootkit: 52 more signatures, fewer false positives, and the perfdata counts rootkits instead of indicators
- statuspal: detects emergency maintenance
Assets:
- sudoers: sudo calls no longer fill the log with PAM session lines
Build, CI/CD:
- the Windows MSI no longer needs an installed Icinga 2 agent
Removed¶
Monitoring Plugins:
- cpu-usage:
--top, useprocs --top - scanrootkit: the Suckit check and the
rootkit_extraperfdata
Tools:
grafana-tool
Fixed¶
Monitoring Plugins:
- about-me: no errors on missing hardware information (#1006)
- cpu-usage: no false 100% on Windows with 64 or more cores (#626)
- deb-updates: no crash when reporting updates
- docker-stats: memory perfdata no longer uses the CPU thresholds
- file-age: copes with files vanishing during the check
- fs-ro: ignores
/run/credentials - keycloak-stats: runs again
- librenms-alerts: reports
WORSE,BETTERandCHANGEDalerts (#882) - logfile: services sharing a logfile no longer interfere, and the check works on Windows (#698, #1035)
- mysql-joins, mysql-traffic: no crash right after a server start
- mysql-memory: no crash with psutil older than 5.3.0
- needs-restarting: reports a pending kernel upgrade on Debian
- notify-host-mail, notify-service-mail: the Icinga logo renders again (#790)
- ntp-*: no more
TypeError - redfish-drives: system-level warnings no longer affect the check (#652)
- rocketchat-stats: no crash when reporting the user count
- service: Windows services with a space in their name match (#921)
- several plugins: logic errors fixed (#1070)
- updates: no crash on Python 3.9
- users: correct TTY count with IPv6 clients (#989)
- valkey-status: TLS works (PR #954, thanks to Claudio Kuenzler)
Build, CI/CD:
- RPM: no conflict with other packages shipping build-id symlinks, such as
azure-cli(#979)
Grafana:
- the Icinga dashboard works with any service name
v2.2.1 - 2025-09-22¶
Fixed¶
Monitoring Plugins:
- ntp-chronyd, ntp-ntpd: no SyntaxError on Python 3.11 (#952)
v2.2.0 - 2025-09-19¶
Added¶
Monitoring Plugins:
- spring-boot-actuator-health: Spring Boot Actuator health (derived from PR #940, thanks to Dominik Riva)
- virustotal-scan-url: URLs flagged by VirusTotal
Build, CI/CD:
- packages for Debian 13 and RHEL 10
Changed¶
Monitoring Plugins:
- cpu-usage: more accurate and faster
- gitlab-health, gitlab-liveness, gitlab-readiness, infomaniak-events: longer default timeouts
- procs: much cheaper on busy Windows servers
- statuspal: a "performance" incident is WARN instead of UNKNOWN
Fixed¶
Monitoring Plugins:
- deb-updates: reports why apt-get fails, and no longer reports OK without the rights to check (#904, #937)
- icinga-topflap-services: no stack trace on empty parameters
- openstack-swift-stat: works with the current python-keystoneclient (#900)
- redis-status, valkey-status:
--ignore-thpworks (#898) - safenet-hsm-state: performance data in the Director basket
- users: no longer reports "no one is logged in" on Ubuntu 24.04 (#919)
Assets:
- SELinux policy: no more denials on D-Bus IPC (#918)
v2.1.1 - 2025-06-20¶
Fixed¶
Icinga Director:
- the Icinga 2 Service Set
v2.1.0 - 2025-06-20¶
Added¶
Monitoring Plugins:
- icinga-version: Icinga end of life
Icinga Director:
- Icinga 2 Service Set
Changed¶
Monitoring Plugins:
- matomo-version: uses the EOL library,
--cache-expireis deprecated
Fixed¶
Monitoring Plugins:
- disk-usage: copes with an inaccessible disk (#792)
- updates: no more "The syntax of the command is incorrect."
Icinga Director:
- nextcloud-app-update.timer unit states
v2.0.0 - 2025-06-06¶
Breaking Changes¶
Build, CI/CD:
- Linux: the packages ship the source code instead of binaries and need Python 3.9 or newer on the host
- Windows: only checks for local resources are compiled, remote checks are meant to run on Linux
Icinga Director:
- the checks no longer compiled for Windows and the legacy commands are gone from the Windows configuration
Added¶
Monitoring Plugins:
- atlassian-statuspage: incidents on an Atlassian Statuspage
- deb-updates: pending updates on
apt-getsystems - kubectl-get-pods: health of Kubernetes pods
- rpm-updates: pending updates and their advisories
- valkey-status: Valkey server statistics
- valkey-version: Valkey end of life
Changed¶
Monitoring Plugins:
- about-me: detects Valkey and the display server
- csv-values: copes with omitted
--warning-queryand--critical-query - icinga-topflap-services: the default warning level rises from 5 to 7
- php-status: bz2 and curl are no longer expected by default
- redfish-sel: supports Supermicro (#866)
Assets:
- sudoers: the command alias is prefixed to avoid conflicts (#880)
Fixed¶
Monitoring Plugins:
- by-ssh: no traceback on "permission denied"
- icinga-topflap-services: no UNKNOWN while Icinga DB synchronizes
- needs-restarting: works again
- ping: "10 received" is no longer read as "0 received" (#860)
- snmp: special characters in SNMPv3 passwords work (#886)
v1.2.0.11 - 2025-03-13¶
Breaking Changes¶
Monitoring Plugins:
- the source-code variant requires Python 3.9 or newer
- jitsi-videobridge-stats:
--warning/--criticalare gone, the check always returns OK (PR #780, thanks to SnejPro)
Notification Plugins:
- notify-*-rocketchat-telegram: the Telegram functionality and the
-telegramsuffix are gone
Icinga Director:
- the Tarifpool-v2 Service Set is removed
Build, CI/CD:
- semantic versioning replaces calendar versioning, starting at
v1.0.0.0
Added¶
Monitoring Plugins:
- new checks: graylog-version, hin-status, icinga-topflap-services, keycloak-memory-usage, keycloak-stats, mastodon-version, moodle-version, openvpn-version, scanrootkit, statusiq, uptimerobot, whmcs-status
Icinga Director:
- new Service Sets: Debian 12 (Cloud Image), IcingaDB, Mastodon, Moodle, networking, rsyslog, Ubuntu 24, WHMCS
Build, CI/CD:
- packages for ARM (#702)
Changed¶
Monitoring Plugins:
- about-me: more accurate VM birth date, detects Mastodon, Moodle and WHMCS
- dhcp-scope-usage: ignores PercentageInUse fractions
- disk-io: supports Windows again
- fs-inodes: checks inode usage per real disk,
--mountis deprecated - infomaniak-events: returns CRIT on critical events
- keycloak-version: reads the version over the REST API (#748)
- librenms-alerts, librenms-health: show non-OK entries only by default
- mysql-thread-cache: measures the hit rate only after one hour of uptime
- nextcloud-security-scan: handles errors from scan.nextcloud.com
- nodebb-stats: "Last user" no longer reports the check's own account (#536)
- openstack-nova-list: no longer needs keystoneauth and keystoneclient
- rocketchat-version: uses the EOL library,
--cache-expireis deprecated - uptime: reports downtime (#191)
Icinga Director:
- the Windows plugins move to
c:\Program Files\icinga2\sbin\linuxfabrik, dmesg uses sudo, and the Debian Service Sets watch/var/log/syslog
Build, CI/CD:
- Windows ships as an MSI package
Fixed¶
Monitoring Plugins:
- about-me: detects expanded RAM (#757)
- apache-httpd-status: works with mod_md enabled (#783)
- dhcp-relayed: binds to all network interfaces
- disk-io: no UnboundLocalError (#777)
- docker-stats:
--always-okworks, and0Bno longer crashes the check (#776, #839) - fortios-network-io: reads its local database again
- needs-restarting: works under the nagios user (#799)
- redfish-sel: no UnboundLocalError (#779)
- service:
--starttypeworks - snmp: no
IndexErroron some device CSV files - strongswan-connections: works with AES-GCM (#806)
- swap-usage: no ProcessLookupError
Removed¶
Build, CI/CD:
- packages for Debian 10, RHEL 7 and Ubuntu 18.04
2024060401 - 2024-06-04¶
Added¶
Monitoring Plugins:
- mysql-query: check running an arbitrary query against a MySQL/MariaDB server
Build, CI/CD:
- packages for Ubuntu 24.04
2024052901 - 2024-05-29¶
Breaking Changes¶
Monitoring Plugins:
- disk-io: rewritten, with new parameters. The perfdata "throughput" is renamed to "bandwidth", only mounted disks are considered, and dm-* device names are translated (#709, #708, #676)
- file-size: the thresholds require a size qualifier,
--warning=10Kinstead of--warning=10000 - journald-query: pattern matching is always case-sensitive (#745)
- librenms-alerts, librenms-health: rewritten to read from the LibreNMS database, with new parameters
- php-fpm: the
--*-max-childrenparameters are gone, because php-fpm's "max children reached" is either 0 or 1 - snmp: update your device CSV files, two more columns are required (#481)
- uptime: warns about recent reboots, and the thresholds require a time qualifier,
--warning=180Dinstead of--warning=180(#722)
Notification Plugins:
- all notification plugins are installed in
/usr/lib64/nagios/plugins/notifications/, otherwise installing both packages at once fails (#726)
Icinga Director:
- the predefined "Journald Query" definitions are gone, single services turned out to be more useful
- many service templates and service set services are less critical by default. Check the ones that matter to you and raise them again
Added¶
Monitoring Plugins:
- new checks: composer-version, dhcp-relayed (a port of check_dhcp_relayed), mediawiki-version
Icinga Director:
- TuneD Service Set, therefore removed from all "OS - RHEL" service sets
Changed¶
Monitoring Plugins:
- about-me: detects non-default software, UDP ports, hardware and much more
- deb-lastactivity: WARNs when the last modified timestamp is missing for one or more packages (#743)
- file-size: supports Nagios ranges for
--warningand--critical(PR #735, thanks to djmcd89) - fs-ro:
/dev/loopis ignored by default, and the output is easier to read (PR #729, PR #730, thanks to Konrad Bucheli) - journald-query: the hard-coded
--bootis gone from the query - librenms-version: reads from the local SQLite database
- mysql-memory: enhanced output, threshold at 95%
- ntp-chronyd: hints at the configuration when no NTP server is being used
- swap-usage: reports the top 3 processes causing the usage (Linux only)
Icinga Director:
- "Starface Java Status" is renamed to "Starface Java Memory Usage", and systemd-units-failed ignores session-c*.scope by default
Fixed¶
Monitoring Plugins:
- about-me: no longer throws an exception for openvas, and a pipe in the output no longer breaks the perfdata (#741, #749)
- csv-values: a pipe in the data is no longer read as the delimiter to the perfdata (#727)
- infomaniak-events: no longer aborts with an UnboundLocalError
- nextcloud-stats: no longer aborts with
KeyError: apps(#731) - ntp-ntpd: unpacks the
ntpq -pvalues correctly (PR #758, thanks to Leo Pempera) - ntp-w32tm: no longer aborts with an UnboundLocalError
Icinga Director:
- corrected the "FreeIPA Server Service Set" definition
2023112901 - 2023-11-29¶
Breaking Changes¶
Notification Plugins:
- the generated URLs point at Icinga DB Web instead of the old IcingaWeb2 Monitoring Module (#643)
Added¶
Monitoring Plugins:
- new checks: apache-solr-version, deb-lastactivity, gitlab-health, gitlab-liveness, gitlab-readiness, gitlab-version, ntp-w32tm, openjdk-redhat-version, openstack-nova-list, postgresql-version, python-version, redis-version, statuspal (#670, #629, PR #710, thanks to Yannic Schüpbach)
Icinga Director:
- Apache Solr Service Set, Debian 12 (Bookworm) coverage including deb-lastactivity
Grafana:
- dashboards for mysql-connections and mysql-memory
Changed¶
Monitoring Plugins:
- *-version: the version data is fetched from endoflife.date first and falls back to the bundled data, with an EOL offset date and optional warnings on a new major, minor or patch release (#680)
- about-me: detects ncdu and yarn, and shows systemd timers with their next runtime
- cpu-usage: excludes the "System Idle Process" from the Windows top 3 list
- disk-smart: skips unsupported disks (#672)
- fortios-firewall-stats: runs when a FortiOS user has only IPv4 or only IPv6 (PR #719, thanks to Pierrot la menace)
- mysql-aria: no longer WARNs on a low
pct_aria_keys_from_mem - mysql-connections: reports and warns on the current usage instead of the peak usage
- mysql-logfile: an empty logfile is OK instead of UNKNOWN, and the auto-configuration stops when
--server-logis given (PR #716, thanks to Eric Esser) - php-version: checks several installed PHP versions (#694)
- qts-*: tested against QuTScloud 4.5.6, 5.0.1 and 5.1, and qts-version no longer reports "up to date" when new firmware is available (#692)
- rocketchat-stats: reports the values Rocket.Chat added (#151)
- uptime: also reports the last reboot time (#190)
Assets:
- sudoers: the command calls are no longer logged
Fixed¶
Monitoring Plugins:
- csv-values: the header is no longer included in the data despite
--skip-header(#706) - journald-query: the perfdata is named "journald-query" instead of "sudo journald-query"
- path-rw-test: uses a unique filename to avoid a race condition (#283)
- qts-disk-smart: works again after a QTS update (#696)
- swap-usage: no longer aborts with
PdhAddEnglishCounterW failed
2023051201 - 2023-05-12¶
Breaking Changes¶
Monitoring Plugins:
- all Python 2 based plugins and libraries are gone, and the "3" suffix is dropped from the Python 3 ones (#589)
- the repository moves to a new directory structure (#350)
- disk-usage:
--ignoreis dropped in favour of including mount points and file systems (#662) - keycloak-version, php-version, wordpress-version: simplified, no longer care about patch levels, no longer need internet access, and dropped some parameters
Assets:
- sudoers: simplified (#651)
Added¶
Monitoring Plugins:
- new checks: apache-httpd-version, by-ssh, cometsystem, fedora-version, githubstatus, grafana-version, mysql-version, network-io, openstack-swift-stat, postfix-version, rhel-version, safenet-hsm-state (#619, PR #648, PR #650, thanks to Dominik Riva)
Grafana:
- new and updated panels, plus a dashboard for the built-in icinga command (#577)
Changed¶
Monitoring Plugins:
- apache-httpd-status: the
ReqPerSec,BytesPerSec,BytesPerReqandDurationPerReqperfdata is gone, the values were wrong - disk-io:
--ignoreignores all disks starting with the given value, and the top 3 I/O processes moved here (#285) - disk-usage: mount points and file systems can be included, absolute values are allowed for the thresholds, and the table also shows "free" (#114, #482, #662)
- fortios-version: simplified, returns the version information in the perfdata
- journald-query: the default
--sincedrops from 24h to 8h - kemp-services: displays the original status of every Virtual Service (#654)
- nextcloud-version: simplified, no longer needs internet access
- php-fpm-status: the meaningless
req per secperfdata is gone - php-status:
monitoring.phpmoved, and one perfdata item is renamed tophp-opcache-memory_usage-current_wasted-percentage - restic-snapshots: shorter output,
--lengthyfor the full table, and the default grouping changes to 'host,paths'
Fixed¶
Monitoring Plugins:
- disk-smart: no longer aborts with
KeyError: 'serial_number'(#659) - disk-usage: works with current psutil (#663)
- file-age: works with current psutil on SMB shares (#665)
- kemp-services: the credentials are converted correctly (#653)
Removed¶
Monitoring Plugins:
- top3-processes-which-caused-the-most-io, moved into disk-io
2023030801 - 2023-03-08¶
Breaking Changes¶
Monitoring Plugins:
- journald-query:
--grepand--case-sensitiveare replaced by--ignore-regex, and the check comes with a new filter (#641) - journald-usage:
--warningswitches from MiB to GiB - mysql-*:
--hostname,--password,--portand--usernameare gone, all checks authenticate through an option file - pip-updates, redis-status, service, veeam-status: rewritten or extended, all four come with new parameters (#646, #623, #630)
Icinga Director:
- removed Service Sets: "OS - RHEL 7 Basic (Hardware)", all six oVirt sets, "PostgreSQL 9.6", plus the duplicate tags "redhat7" and "redhat8"
Added¶
Monitoring Plugins:
- new checks: crypto-policy, csv-values, grassfish-licenses, grassfish-players, grassfish-screens, infomaniak-events, journald-query, journald-usage, ntp-chronyd, ntp-ntpd, ntp-systemd-timesyncd, restic-check, restic-snapshots, restic-stats, systemd-timedate-status, tuned-profile. The three ntp checks replace ntp-offset (#449)
Notification Plugins:
- notify-host-rocketchat-telegram, notify-host-zoom, notify-service-rocketchat-telegram
Icinga Director:
- new Service Sets: AIDE, Apache for Debian 11, Basic for Debian 11, FreeIPA Server, Grav, Ubuntu 22, UPS (Network UPS Tools), plus debug-shell.service in all RHEL-based Basic Service Sets
Assets:
- SELinux Type Enforcement Policies, and sudoers files for Alma 9, RHEL 9, Rocky 9, Fedora 37, Oracle 7, Oracle 8 and Oracle 9 (#627)
Changed¶
Monitoring Plugins:
- about-me: rewritten, recommends tags for the Icinga Director basket with
--tags, detects restic and Snap, reports maker and model, and the external IP lookup is configurable and off by default (#637, #645) - disk-usage: the state moves into the usage column
- fs-ro: squashfs and ramfs are excluded (#412, #617)
- infomaniak-swiss-backup-*: adapted to the new API version
- mysql-connections:
--ignore-name-resolution(#631) - mysql-user-security: ignores the mysql.sys and mariadb.sys users
- network-connections: alerts above a configurable number of connections (#621)
- php-status: the URL to monitoring.php is optional, and startup, config and module errors are reported more clearly
- redis-status: no longer warns on "Peak memory"
- service: checks several Windows services at once (#609)
Icinga Director:
- the MariaDB/MySQL service set is split into InnoDB, Metrics, Replication, Schemas, Security and a baseline set
- the RHEL and Fedora sets gain TuneD Profile and Crypto Policy, every set with a systemd service gains a matching Journald Query and Systemd TimeDate Status, and notifications are enabled only for critical hardware-related services
Fixed¶
Monitoring Plugins:
- disk-usage: CDFS is ignored by default (#632)
- docker-stats: the container name in the perfdata is shortened as intended (#600)
- file-age: new files are no longer reported critical because of a negative modification time (#618)
- infomaniak-swiss-backup-devices: no longer aborts with a TypeError
- librenms-version: no longer aborts with
KeyError: 'mysql_ver'(#602) - matomo-reporting:
--metricreturns the one metric asked for (#603) - nextcloud-stats: no longer aborts on a missing ALWAYS_OK attribute (#640)
- ping: no longer aborts because
ping -twas handed a float (#628) - rpm-lastactivity: no longer aborts with a ValueError (#616)
- updates: no longer returns a PowerShell error on Windows behind a closed firewall (#633)
Removed¶
Monitoring Plugins:
- ntp-offset, split into ntp-chronyd, ntp-ntpd and ntp-systemd-timesyncd (#449)
- all plugins: the code for self-handling Python virtual environments (#543)
Icinga Director:
- DiagTrack, Windows telemetry, is gone from the Windows Service Sets
2022072001 - 2022-07-20¶
Breaking Changes¶
Monitoring Plugins:
- wildfly-memory-pool-usage:
--warningand--criticalare gone, they are not needed any more (#563)
Added¶
Monitoring Plugins:
- new checks: diacos, infomaniak-swiss-backup-devices, infomaniak-swiss-backup-products, strongswan-connections, xml, the whole mysql-* family (aria, binlog-cache, connections, database-metrics, innodb-buffer-pool-instances, innodb-buffer-pool-size, innodb-log-waits, joins, logfile, memory, open-files, perf-metrics, replica-status, slow-queries, sorts, storage-systems, system, table-cache, table-definition-cache, table-indexes, temp-tables, thread-cache, traffic, user-security) and the nodebb-* family (cache, database, errors, events, groups, info, users) (PR #567, PR #583, thanks to Dominik Riva)
Icinga Director:
- duplicity Service Set, strongSwan Service Set
Assets:
- sudoers files for Fedora 35 and Fedora 36
Changed¶
Monitoring Plugins:
- about-me: reports birthdate, boot mode, listening ports, the active tuned profile and the key features of the machine, and detects AIDE, certbot, acme.sh, gpg, mod_security and swanctl
- all checks using SQLite databases: more unique database names (#333)
- cpu-usage: the "nice" percentage is subtracted from the thresholds (#550)
- dhcp-scope-usage: parses PercentageInUse locale-aware (PR #551)
- disk-smart: runs on Windows, and excludes zfs volumes (PR #539, PR #553)
- disk-usage: the first output line no longer says "OK" while the check is critical (#545)
- docker-info: raises CRIT on a return code other than 0 (#569)
- docker-stats: better handling of container names (#586)
- file-age: shorter message and better perfdata labels (#559, PR #544)
- ipmi-sel: the events are ordered differently (#558)
- needs-restarting: works on Debian Buster and Bullseye (#572)
- php-status: different handling of default values, and
--devsuppresses the warnings on display_errors and display_startup_errors (#461, #540) - qts-*: the default connect timeout rises from 3 to 6 seconds
- systemd-units-failed:
--ignoreaccepts wildcards (#542)
Icinga Director:
- longer check intervals for the Windows services
Fixed¶
Monitoring Plugins:
- file-count: no longer aborts with
KeyError: 'lib'(#591) - fortios-memory-usage: works on Python 3 (PR #599)
- keycloak-version: no longer aborts on a missing match (#555)
- logfile: no longer runs into "Database locked" and UNKNOWN under heavy use on one host (#578)
- xca-cert: checks the expiry date again
Removed¶
Monitoring Plugins:
- mysql-stats, nodebb-stats, nodebb-status
Icinga Director:
- gpsvc on Windows
2022030201 - 2022-03-02¶
This is a "we migrated everything from GitLab to GitHub, but had to adjust many details afterwards" version. In terms of source code, nothing has changed compared to 2022022801, just a bunch of links in source code comments and READMEs.
2022022801 - 2022-02-28¶
Breaking Changes¶
- this is the last release including bugfixes for the Python 2 variant of all checks
- the project moved from our self-hosted GitLab to a public repo on GitHub. The branches "master" and "develop" are gone, releases are built from tags on "main", and all commit hashes changed with the removal of the binaries
- the checks compiled for Windows moved to the download server
Added¶
Monitoring Plugins:
- dhcp-scope-usage: IPv4 scope usage of a Windows DHCP server, locally via PowerShell or remotely via WinRM
- huawei-dorado-*: backup power modules, controller, disks, enclosures, fans, interfaces, power, attached hosts, the system itself and the HyperMetro domain and pairing information of a Huawei OceanStor Dorado storage system
- redfish-drives, redfish-sel, redfish-sensor: drives, system event logs and sensor data of a Redfish-based BMC
Notification Plugins:
- notify via Zoom, notify via e-mail
Icinga Director:
- new Service Sets: acme.sh, Active Directory Certificate Services, Active Directory Domain Services, Active Directory Federation Services, Active Directory Lightweight Directory Services, DHCP Server, DHCP Server Failover Feature, DNS Server, Duplicati, Huawei Dorado, Redfish, Redfish no agent, Veeam Backup & Replication, Web Server (IIS), Windows Basic extended, Windows Defender Antivirus Service
Assets:
- sudoers for Rocky 8 and openSUSE Leap 15
Changed¶
Monitoring Plugins:
- about-me: reports virtualisation, detects ownCloud and an alternate Nextcloud path, and runs even when psutil is missing (#480, #512, #514)
- librenms-alerts, librenms-health: more filtering parameters, among them
--device-group - nginx-status: prints human-readable total values (#520)
- php-status: hints when it is not running with sudo (#459)
- redis-status: supports Redis 3.0, is more tolerant about defragmentation, warns only below a 10% cache hit rate, and warns on a bad OS configuration (#425, #428, #490, #510)
- rocketchat-stats: rocket.chat is renamed to rocketchat (#335)
- swap-usage: no longer displays "swapped in" and "swapped out" on Windows (#454)
- veeam-status:
--usernameand--passwordare mandatory (#499) - wildfly-deployment-status: the deployment can be limited by name (#497)
Icinga Director:
- the Huawei service names and the Windows variants are adapted, notifications are enabled for the Redfish checks, the LibreNMS services are split by type, and getent gets a 30 second timeout (#455)
Fixed¶
Monitoring Plugins:
- about-me: no longer aborts on a VMware hypervisor check, an index error or a missing psutil attribute (#438, #443, #513)
- apache-httpd-status: no longer aborts on Ubuntu 16.04 (#436)
- borgbackup: no longer aborts with an AttributeError (#430)
- disk-smart: disk names such as sdda and sdab are checked (#487)
- file-age: files dated in the future are handled, and the Windows variant no longer crashes on a glob wildcard (#478, #494)
- fs-xfs-stats: handles an I/O error while reading /proc/fs/xfs/stat (#445)
- jitsi-videobridge-status: no longer aborts with a TypeError (#527)
- librenms-health: no longer times out on too many values (#365)
- nextcloud-stats: no longer aborts on encoding errors, and the DB size is no longer reported in YiB (#463, #517, #531)
- nginx-status: the perfdata is correct (#440)
- ntp-offset: no longer regularly UNKNOWN when used with chrony (#71)
- php-status: handles a missing display_startup_errors (#434)
- php-version: no longer warns about a patch release that is not newer (#435)
- procs: no longer aborts on Windows, no longer reports "oldest proc created 52Y 1M ago", and reports a missing process as missing (#453, #488, #506)
- redis-status: no longer reports a false somaxconn warning, detects Redis forced into swap, and no longer warns about the password on the command line (#450, #458, #486)
- swap-usage: no longer aborts with an UnboundLocalError (#456)
- systemd-unit: failed units are printed with the correct columns on Fedora, and an empty UnitFileState is handled (#328, #509)
- users: no longer aborts on a decoding error on Windows, and a pipe symbol in the "WHAT" column no longer breaks the output (#17, #451)
- veeam-status: no longer aborts with a ValueError (#45)
- the Windows builds ship the required third-party Python modules again (#504)
Grafana:
- dns: the panels no longer divide the query time by 1000, and the fail2ban panel no longer lists "Banned IPs" twice (#139, #453)
Icinga Director:
- corrected the GUIDs in all-the-rest.json
Removed¶
Icinga Director:
- SysMain and TimeBrokerSvc are gone from the Windows Service Set, and getent from the basic Service Sets (#427, #446)
2021101401 - 2021-10-14¶
Added¶
Monitoring Plugins:
- all checks are ported to Python 3 (suffixed by
3), most of them are also available on Windows, and all of them run on Rocky and Alma Linux - new checks: jitsi-videobridge-stats, jitsi-videobridge-status, nodebb-stats, nodebb-status, nodebb-version, redis, sap-open-concur, veeam-status, and the starface-* family (account, database and peer statistics, overall, backup and channel status, Java memory usage)
Event Plugins:
- cloudflare-security-level
Icinga Director:
- the Windows Basic Service Set gains disk-io, dns, swap-usage and top3-processes-which-caused-the-most-io
Changed¶
Monitoring Plugins:
- about-me: reports much more inventory, among it interfaces, the systemd default target, timers, enabled units, mounts, automounts, non-default users and crontabs, plus GCC, GitLab, OpenVPN, Veeam and vsftpd detection
- apache-httpd-status: calculates ReqPerSec, BytesPerSec, BytesPerReq and DurationPerReq over Apache's uptime, and prints the worker percentage in the table
- dmesg: the output is capped at ten lines, and more messages are ignored by default (#254, #338)
- file-ownership: also checks /tmp/linuxfabrik-plugin-cache.db, with corrected defaults for Debian, SLES and Ubuntu (#294, #317, #332, #356)
- getent: also prints the response (#297)
- php-*: report more, which needs the new
monitoring.phpinstalled - php-status: the cache hit rate check is optional, "simplexml" is no longer a default module, and config and module errors are clearer (#267, #284, #303)
- php-version: checks major and minor by default rather than the patch level, and tests against the package manager (#253, #304)
- procs: counts more accurately, and can alert on specific processes (#355)
- systemd-unit:
--unitfilestateaccepts None to disable the unit file state check (#299) - wildfly-gc-status, wildfly-memory-pool-usage: higher defaults for
avr_gc_time, and "PS_Survivor_Space" no longer alerts (#286, #307)
Icinga Director:
- the command definitions are provided through the basket (#301)
Fixed¶
Monitoring Plugins:
- about-me: shows all disks, no longer reports loolwsd when it is not installed, and no longer aborts on unpacking (#281, #370, #372)
- apache-httpd-status: no longer aborts on an unsupported operand type (#323)
- disk-io: negative rate differences after a reboot are handled (#312)
- dmesg: no longer counts one line too many (#331)
- file-age: negative times are handled correctly (#188)
- getent: no longer aborts on a decoding error (#367)
- mydumper-version: copes with a version such as "v0.10.7-2" (#318)
- network-port-tcp: no longer aborts with a NameError (#298)
- php-status: monitoring.php runs on PHP 7.2, and OPcache is no longer reported as missing when monitoring.php is not used (#289, #290, #324)
- php-version: no longer aborts on a Debian package version (#293)
- procs: the counting in the output is correct (#357)
- qts-temperatures: no longer aborts with a traceback (#360)
- service: reports the right state when a service is running but is not supposed to be (#336)
- systemd-unit: an empty UnitFileState is handled (#292)
Removed¶
Monitoring Plugins:
- fah-stats, hostname and all Atlassian checks
2021061501 - 2021-06-15¶
Added¶
Monitoring Plugins:
- half of the checks are ported to Python 3 (suffixed by
3), 17 of them are available on Windows - new checks: docker-info, docker-stats, fs-xfs-stats, haproxy-status, librenms-alerts, librenms-version, logfile, metabase-stats, mod-qos-stats, mydumper-version, nginx-status, onlyoffice-stats, path-rw-test, php-fpm-ping, php-fpm-status, php-status, php-version, pip-updates, snmp, and the wildfly-* family (deployment status, garbage collector status, memory and memory pool usage, server status, thread usage, uptime, XA and non-XA datasource statistics)
- the human-readable units in the output are more precise: "MiB" always means Mebibyte, "M" means Month while "m" means Minute, and the SI symbol "G" means Billion
- the performance data names move towards Prometheus compatibility, starting with fs-xfs-stats and nginx-status
Changed¶
Monitoring Plugins:
- about-me: detects Django, LibreNMS, mydumper, Nikto, OpenSSL, OpenVAS, tmate and more, reports the local and public IP address, and ignores zram devices (#227, #256)
- cpu-usage, file-descriptors, memory-usage: the three "Top 3" checks are merged into them (#246, #247, #248)
- disk-io: determines the maximum possible disk throughput automatically, and the separate "State" column is gone (#279)
- dmesg: uses
--ctimeinstead of--reltime, caps the output at ten lines, and gained a severity parameter (#115, #238, #254) - feed: no longer fetches items dated in the future, strips HTML from the content, and falls back to the "content" field of an Atom feed (#95, #206, #207)
- file-*: deal directly with SMB/CIFS shares
- file-ownership: prints a table, and the default list follows the CIS CentOS standard (#231, #233)
- fortios-*: the port can be specified, and the password is HTTP-encoded (#186, #187)
- ipmi-*: can connect remotely to Supermicro IPMI, HPE iLO and DELL iDRAC (#168, #169)
- nextcloud-version: takes the Apache user from the owner of config/config.php, and handles the Enterprise channel (#142, #225)
- procs: the filters for username, process name and arguments are case-insensitive, the used filter is shown, and memory usage is always in the perfdata (#261, #263, #264)
- wildfly-gc-status: collection time and count are reported as continuous counters (#185)
Fixed¶
Monitoring Plugins:
- about-me: disk sizes show up on CentOS (#259)
- apache-httpd-status: copes with HTML pages containing "::" (#199)
- disk-io: a negative RW5 is clamped to 0 (#265)
- disk-smart: ignores zram devices, and no longer aborts with a SyntaxError (#220, #221)
- feed: no longer runs into the 10s plugin timeout (#83)
- nextcloud-stats: num_users no longer counts every user who ever existed (#224)
- procs: checking processes by CPU usage is correct, and several tracebacks are gone (#162, #166, #260)
- users: no longer aborts on a decoding error on Windows (#201)
Removed¶
Monitoring Plugins:
- fs-file-usage, replaced by file-descriptors (#234)
- three of the four "Top 3" checks, merged into cpu-usage, file-descriptors and memory-usage
2021021701 - 2021-02-17¶
Fixed¶
Monitoring Plugins:
- the virtualenv is activated even when a plugin is called by an absolute path (#154)
2021021601 - 2021-02-16¶
Added¶
Monitoring Plugins:
- support for running the plugins in a virtual environment
Changed¶
Monitoring Plugins:
- file-age, file-count, file-size: support SMB
- nextcloud-version: returns UNKNOWN when the update server is unavailable, with a longer timeout (#147, #148)
- procs: thresholds for CPU and memory
- users: the missing perfdata on Windows
Fixed¶
Monitoring Plugins:
- json: renamed to json-values, it collided with the official json library
- pip-version: the output message is no longer mixed up
2020122401 - 2020-12-24¶
Added¶
Monitoring Plugins:
- new checks: dummy, file-count, json
- Windows builds for cpu-usage, disk-usage, dummy, file-age, file-count, file-size, json, memory-usage, network-connections, procs, scheduled-task, service, updates, uptime and users
Changed¶
Monitoring Plugins:
- file-age, file-size: support globbing to select several files
Fixed¶
Monitoring Plugins:
- users: the count on Windows includes disconnected users
2020112001 - 2020-11-20¶
Changed¶
Monitoring Plugins:
- systemd-unit: knows more states
2020111901 - 2020-11-19¶
Fixed¶
Monitoring Plugins:
- ntp-offset: no longer errors on a server without NTP (#138)
2020111801 - 2020-11-18¶
Added¶
Assets:
- sudoers for Debian 9 and 10
Fixed¶
Monitoring Plugins:
2020102301 - 2020-10-23¶
Breaking Changes¶
- the repository is restructured, and the first Windows-compatible Python 3 check plugins arrive
Added¶
Monitoring Plugins:
- new checks: atlassian-confluence-version, atlassian-jira-version, keycloak-version, pip-version, wordpress-version, and the qts-* family (cpu-usage, disk-smart, memory-usage, temperatures, uptime, version)
- Windows builds for cpu-usage, disk-usage, file-age, file-size, memory-usage, network-connections, procs, scheduled-task, service, updates, uptime and users
Tools:
- grafana-tool, a utility to generate Grafana dashboards
Changed¶
Monitoring Plugins:
- borgbackup: the expected string in the logfile changes from rc to retc
- feed:
--no-icinga-callbackis replaced by--icinga-callback
Assets:
- the sudoers are unified into one file per OS, in the assets/sudoers folder
Fixed¶
Monitoring Plugins:
- feed: no longer aborts with a traceback (#107)
- memory-usage: prints the top 3 memory consuming processes on WARN and CRIT (#108)
- ntp-offset: supports systemd-timesyncd (#90)
- openvpn-client-list: the output is a table (#19)
- qts-version: no longer reports "None" after an update (#112)
- xca-cert: lists all checked certificates with common name, CA, serial and expiry date (#65)
2020061901 - 2020-06-19¶
Added¶
Monitoring Plugins:
- network-bonding
Fixed¶
Monitoring Plugins:
- nextcloud-version: no longer aborts with an AttributeError (#105)
2020052801 - 2020-05-28¶
Added¶
Monitoring Plugins:
- new checks: fs-ro, kemp-services, matomo-reporting, matomo-version, sensors-battery, sensors-fans, sensors-temperatures, systemd-units-failed, and the fortios-* family (cpu-usage, firewall-stats, ha-stats, memory-usage, network-io, sensor, version)
Changed¶
Monitoring Plugins:
- most of the checks also run on Ubuntu Server 16 and newer
- any token and password URL parameter is printed with asterisks on a stack trace
- all checks calling shell commands force English output even when the system locale differs
Fixed¶
Monitoring Plugins:
- apache-httpd-status: no longer aborts on Ubuntu 16, and reports a malformed server-info (#97, #101)
- disk-io: no longer counts loop devices on Ubuntu 20, and no longer aborts on Ubuntu 16 (#87, #98)
- disk-smart: no longer aborts when not running on hardware (#82)
- disk-usage: ignores snap and iso9660 devices (#88, #100)
- mailq: works with Exim (#93)
- nextcloud-version: no longer returns UNKNOWN when the update server is unavailable (#99)
- procs: the total process count in the perfdata is no longer always 0 (#96)
2020042001 - 2020-04-20¶
Added¶
Monitoring Plugins:
- new checks: dns, fah-stats
- most of the checks also run on Ubuntu
Fixed¶
Monitoring Plugins:
- about-me: reports the details of NVMe disks (#89)
- nextcloud-security-scan: no longer aborts on a missing urllib (#91)
- ping: no duplicate output (#84)
2020041501 - 2020-04-15¶
Added¶
Monitoring Plugins:
- new checks: getent, nextcloud-version, ping, rocket.chat-version
Removed¶
Monitoring Plugins:
- docker-info, docker-container, network-io, redis and xca-cert, to be rewritten from scratch
2020031201 - 2020-03-12¶
Added¶
Monitoring Plugins:
- feed
Changed¶
Monitoring Plugins:
- cpu-usage: adjusted to changes in psutil
- dmesg: a longer ignore list
- systemd-unit: improved output
2020022801 - 2020-02-28¶
Initial release for the general public.