In that case, why not use the timeout stuff from the services? You could probably override the service to set a timeout for its running with [Unit] JobRunningTimeoutSec=300 JobTimeoutAction=reboot-force
So after starting you get up to 5 minutes of it running and then it should reboot. Probably the JobTimeoutAction is not needed as I'm sure it will reboot if the emergency.service fails, but better safe than sorry :) That gives you the best of both worlds. You get some time to debug and if not it will auto reboot after a bit and maybe fix itself :D On Wed, Jun 12, 2024 at 2:21 AM Mark Zhang (NSB) <mark.zh...@nokia-sbell.com> wrote: > Hello Itxaka, > > > > Thanks for the suggestion. > > If we override with reboot, that may impact the ability of debug in case > encountered real issue. > > > > Thanks, > > Mark > > > > *From:* Itxaka Serrano Garcia <itxaka.gar...@spectrocloud.com> > *Sent:* 2024年6月11日 16:08 > *To:* Mark Zhang (NSB) <mark.zh...@nokia-sbell.com> > *Cc:* systemd-devel@lists.freedesktop.org > *Subject:* [External] Re: [systemd-devel] How to set the timeout of > emergency mode > > > > > > *CAUTION:* This is an external email. Please be very careful when > clicking links or opening attachments. See http://nok.it/nsb for > additional information. > > > > > > Maybe you could override the emergency.service to reboot automatically? > Either on the ExecStartPost or even earlier if you dont want to even show > the logging, on ExecStartPre? > > > > I think that could work. > > > > On Tue, Jun 11, 2024 at 5:19 AM Mark Zhang (NSB) < > mark.zh...@nokia-sbell.com> wrote: > > Hello Experts, > > > > Consult one question about how to set the timeout of emergency mode? > > > > Scenario: > > We have encountered one problem of Openstack host, and the host was up at > the end, and all things are good from host level after that. > > But the VM stuck in the emergency mode forever. If we input ctrl+D, the VM > could reboot normally again. > > > > So we would like to consult, if any method could avoid to stuck in the > emergency mode forever? Be better, there is some timeout setting for > emergency mode. Then the VM could auto-recover by itself for such scenario. > > > > Thanks, > > Mark > > > >
