On Sat, May 01, 2004 at 12:56:47PM +0300, Shachar Shemesh wrote: > Hi all, > > Does anyone know of a mechanism that will allow me to have a kernel > automatically reboot on panic? > > When remote upgrading a kernel, the most common problems have to do with > either initrd creation or some other form of not finding the root > partition. Once the root partition is mounted, it is very rare to have > any problems after that point. > > Now, LILO has this wonderful mechanism, where you can have a certain > kernel for your next boot, and a different kernel for all subsequent > reboots. This is great as you can do two commands before the crucial reboot: > lilo -D failsafe > lilo -R newkernel > > and the new kernel will boot once, and the failsafe kernel will boot the > rest of the times. This still leaves one problem - how do you recover if > the new kernel did not start properly? Today, I have to call the > co-location ISP, convince them that the machine is indeed down, and ask > them to reboot it. I then try to fix the problem, and reboot again, > which may cause the machine to hang again, and so on and so forth. > > What I'm looking for is a kernel command option (I can put it in > lilo.conf) that tells it "if you panic, just reboot". This will save me > the phone call in the middle, and yeild much more robust behaviour.
I don't know of any such option. But I figure that a "reboot if rootfs is not available" option is simpler and more suitable. There is also no need for the system to try to sync anything, as typically no real filesystem is mounted at this stage (unless you did something strange in an intrd). -- Tzafrir Cohen +---------------------------+ http://www.technion.ac.il/~tzafrir/ |vim is a mutt's best friend| mailto:[EMAIL PROTECTED] +---------------------------+ ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
