On 08/01/13 at 10:24pm, Baoquan He wrote: > On 07/31/13 at 04:31pm, Harald Hoyer wrote: > > On 07/31/2013 04:29 PM, Vivek Goyal wrote: > > > On Wed, Jul 31, 2013 at 04:25:57PM +0200, Harald Hoyer wrote: > > > > > > Hi Harald, > > > > > > Do you still like the parameter name "rd.action_on_fail". I think a new > > > parameter say, "rd.no_emergecny_shell" might make more sense to reflect > > > what we are doing. > > > > > > If you like it, I can request bao to post a new patch and make appropriate > > > changes. > > > > > > Thanks > > > Vivek > > > > > > > You can add the rd.no_emergency_shell parser in your kdump module and touch the > > marker file. I think that gives you more flexibility for runtime decisions. > > You can just add and remove that file on will. > Hi Vivek, > > I agree with Harald, there's no need to add a parameter into dracut any > more. Because currently only 2 actions are taken, 1 is emergency-shell, > the other is continue. We can just put a file as switch, if the file exists, > all failed action doesn't trigger emergency shell before Kdump, just > continue. > > If you want several more actions, E.g reboot or others, we can put them > into this file. Then in Kdump module, we can read the content and parse. > As for modules before Kdump, they don't feel it. I think this can extend > the 'default action' flexibly. This keep our reliability on dracut at the > least. Is this necessary to define more actions? We can bypass failures (switch off emergency shell) in drauct or systemd and eventually get down to kdump.sh. If the dump fails, "default action" is triggered. Other unrelated failures happening before kdump.sh should be ignored. Thanks WANG Chao -- To unsubscribe from this list: send the line "unsubscribe initramfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html