watchdogd 3.5

watchdogd(8) is an advanced system and process supervisor daemon. It can monitor critical system resources, supervise the heartbeat of processes, record deadline transgressions, and reset the system.

Tags system supervision watchdog
License Apache
State stable

Recent Releases

3.502 Dec 2021 06:25 minor feature: github: change install directory /opt - gt; /tmp Seems /opt is full of stuff so a tree walk takes a loooot of time. Signed-off-by: Joachim Wiberg lt;troglobit@gmail.com gt;.
3.430 Apr 2021 17:43 major bugfix: - Clarify nomenclature: reset cause vs. reset reason - Change layout and formatting of `watchdogctl` status output - Change defaults for supervisor, still disabled by default but now also with priority set to zero by default. This allows running the supervisor in cgroups v2 systems without realtime priority. - Fix problem with plugins being enabled (but incomplete) by default. Now all sections have an `enabled = true false ` setting, and all are disabled by default. You need to uncomment *end* enable. A new command #039;fail #039; has been added to watchdogctl. It can be used with the supervisor script to record the reset cause and do a WDT reset. The reset CAUSE can be forwarded by the script to record the correct (or another) reset cause. Add -p PID to watchdogctl. Works with reset and fail commands. Always warn at startup if driver/WDT does not support safe exit, i.e. quot;magic quot;. Add warning if.conf file cannot be found. Add recorded time of reset to reset cause state file. Omitting critical/reboot level from a checker plugin causes default value of 95 to be set, causing reboot by loadavg plugin. by defaulting to #039;off #039; for checker/monitor critical/reboot level. mismatch in label length between supervised processes and that in wdog_reason_t = increase from 16 to 48 chars. problem disabling the process supervisor at runtime, it always caused a reboot.
3.128 Jun 2018 05:25 minor feature: Supervised processes can now also cause reset if the ACK sequence is wrong when kicking or unsubscribing. Add support for callback script to the process supervisor: script = /path/to/script.sh in the supervisor section enables it. When enabled all action is delegated to the script, which is called as: script.sh supervisor CAUSE PID LABEL. For more information, see the manual for watchdogd.conf. A new command 'fail' has been added to watchdogctl. It can be used with the supervisor script to record the reset cause and do a WDT reset. The reset CAUSE can be forwarded by the script to record the correct (or another) reset cause. Add -p PID to watchdogctl. Works with reset and fail commands. Always warn at startup if driver/WDT does not support safe exit, i.e. "magic ". Add warning if.conf file cannot be found. Add recorded time of reset to reset cause state file. Omitting critical/reboot level from a checker plugin causes default value of 95 to be set, causing reboot by loadavg plugin. by defaulting to 'off' for checker/monitor critical/reboot level. mismatch in label length between supervised processes and that in wdog_reason_t = increase from 16 to 48 chars. problem disabling the process supervisor at runtime, it always caused a reboot.
3.010 Feb 2018 16:08 major feature: Greatly simplified command line interface, advanced features now available in `/etc/watchdogd.conf`. Improved libwdog API for process monitoring. System status monitoring is now optional, with optional reboot on overload. For more details, see the ChangeLog.
3.0-rc121 Nov 2017 04:05 minor feature: Update README with simple API example. Make it possible to run automatic tests as non-root. Add automatic testing of PMON API to Travis. Add Coverity Scan. Silence GNU ar output, has suddenly started warning about ar crus. Only write() to watchdog if descriptor is valid, annoying with watchdog not being properly disabled with wdt_enable(). in wdt_enable() which could possible deref. NULL pointer. Only change enabled state in wdt_enable() if operation is succesful.
3.0-beta113 Apr 2017 03:15 minor feature: Update README with simple API example. Make it possible to run automatic tests as non-root. Add automatic testing of PMON API to Travis. Add Coverity Scan. Silence GNU ar output, has suddenly started warning about ar crus. Only write() to watchdog if descriptor is valid, annoying with watchdog not being properly disabled with wdt_enable(). in wdt_enable() which could possible deref. NULL pointer. Only change enabled state in wdt_enable() if operation is succesful.
2.0.109 Apr 2017 17:32 minor bugfix: