ipmi_watchdog(7)

πŸ“¦ version 3.1.3, in opensuse-leap 16.0

✨ ipmi_watchdog(7) β€” the IPMI watchdog, explained by your favorite terminal gremlin ✨

ok so this is a page from the ClusterTools2 manual set, and we're talking about ipmi_watchdog(7) β€” yes the whole thing is basically "how do I not lose my server at 3am," and honestly? iconic. let's get into it πŸ’…


🌸 NAME 🌸

ipmi_watchdog β€” the IPMI watchdog. that's it, that's the name, she's straightforward and we respect that 🫢


πŸ’– DESCRIPTION πŸ’–

okay bestie, so the kernel module ipmi_watchdog is the one you want on systems that hand you hardware watchdog functionality through an IPMI board. a real example of such a board: the Lenovo Integrated Management Module. lowkey fancy.

and here's the plot twist that makes her different from all the other watchdog modules out there β€” ipmi_watchdog needs a running daemon. like, she does not work solo. she needs her emotional support daemon, no cap 😭


πŸŽ€ EXAMPLES πŸŽ€

prepare IPMI watchdog, 1st try

ok so here's attempt number one, the classic route:

# chkconfig ipmi on

# vi /etc/sysconfig/bootloader
INITRD_MODULES="... ipmi_watchdog"

# vi /etc/modprobe.d/ipmi_watchdog
options ipmi_watchdog nowayout=0 timeout=30 action=reset_power
blacklist iTCO_wdt
blacklist iTCO_vendor_support

# shutdown -r now

like, turn the service on, drop ipmi_watchdog into INITRD_MODULES in /etc/sysconfig/bootloader, write those options and blacklist lines into /etc/modprobe.d/ipmi_watchdog, and then reboot. classic. reboot fixes everything, allegedly 😌

prepare IPMI watchdog, 2nd try

and if that didn't hit, here's the other flavor:

# chkconfig ipmi on

# vi /etc/sysconfig/ipmi
IPMI_WATCHDOG=yes
IPMI_WATCHDOG_OPTIONS="nowayout=0 action=reset_power timeout=30"

same energy, different config file. she's flexible like that πŸ’…

show

and then you wanna check that it actually worked, right? because vibes are not evidence. so:

# rcipmi status

# ipmitool mc watchdog get

# modinfo ipmi_watchdog

# ls -l /dev/watchdog

look at her go. four commands and you know everything. queen behavior πŸ‘‘


πŸ¦‹ FILES πŸ¦‹

  • /dev/watchdog β€” the watchdog device node. the main character herself.
  • /lib/modules/<kernel-version>/kernel/drivers/char/ipmi/ipmi_watchdog.ko β€” the IPMI watchdog module. yes, <kernel-version> is a placeholder, swap in your actual kernel version, i'm begging πŸ₯Ί
  • /etc/init.d/ipmi β€” the IPMI service start script.
  • /etc/sysconfig/ipmi β€” the IPMI related configuration.
  • /etc/modprobe.d/ipmi_watchdog β€” the kernel module configuration.
  • /etc/sysconfig/kernel β€” the related initrd configuration.

πŸ’€ BUGS πŸ’€

feedback is welcome! please mail to feedback@suse.com β€” they genuinely want to hear from you, not me being sarcastic, this is a real invitation 🫢


🌈 SEE ALSO 🌈

go read these, they're all at least a little relevant:


πŸ‘‘ AUTHORS πŸ‘‘

S.Erdenberger, L.Pinne β€” legends, honestly.


(c) 2016-2017 SUSE Linux GmbH, Germany. (c) 2018-2019 SUSE LLC

and the part everyone skips but shouldn't: ipmi_watchdog comes with ABSOLUTELY NO WARRANTY. like, none. zero. don't come for them when it breaks, bestie 😭 for details see the GNU General Public License at http://www.gnu.org/licenses/gpl.html.


ok that's the whole page!! this edition dropped on 01 Nov 2019, courtesy of the ClusterTools2 girlies, and honestly she ate. go forth and watchdog responsibly πŸ’–βœ¨πŸŽ€

πŸ“¦ where this page comes from