On Fri, Feb 01, 2013 at 03:22:49PM +0100, Allmeroth, Robert wrote: > On embedded systems it is sometimes not very useful to call shutdown scripts > or reboot(2) in case of a watchdog failure / failed service (re)starting, > e.g. FS/Flash corruption, FS driver failure, special HW init failure > > - Typically important file systems are read-only on embedded systems so it > would > not harm the system to do a hard reset. > - The processor on which systemd is running on is not necessarily the > PowerController, a call > to reboot(2) would only stop/reboot that node. But other nodes like FPGA, > ASIC, DSP also > need a reset. > > Does it make sense to you to have a StartLimitAction= where the systemd stops > sending > heartbeats to the kernel? (e.g. StartLimitAction = StopKernelHeartbeat > <optional reason>) > In this case the watchdog on the PowerController is able to react very fast > accordingly. > > Of course this will only work if the kernel watchdog module is already > connected to the PowerController. > If not - the only escape is the 'expect watchdog register timer' of the > PowerController. > In general we would like to avoid running into the expect timer because we > have no hint why the system > did not respond.
Have you looked at StartLimitAction=reboot-immediate? This should restart your system immediately without stopping services etc. Michael -- Pengutronix e.K. | | Industrial Linux Solutions | http://www.pengutronix.de/ | Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
