Petter Reinholdtsen wrote: > [Harald Dunkel] >> I just want to move a service out of reach of init and invoke-rc.d . > > Right. And why does it not work to simply disable the service by > replacing start symlinks with stop symlinks in runlevels 2-5? It is > as far as know, the only way to disable a service and get predictable > behaviour. >
/etc/init.d/rc is not supposed to to _stop_ the HA services, either, independent of the run level history. AFAICS setting all symlinks to "K??*" is not sufficient for this. >> The point is, that invoke-rc.d and /etc/init.d/rc treat this 3rd >> choice in an inconsistent way. invoke-rc.d starts the service, if >> the symlink is missing. /etc/init.d/rc doesn't. > > This is what I mean by 'undefined'. The handling of 'no symlink' is > not well defined, and you get unexpected behaviour. :) > From this point of view it is not defined at all. Problem is that there are pretty many runlevel scripts making use if this "undefined" behavior. > Anyway, if I understand this request correctly, you want us to look > into how invoke-rc.d treat the lack of a symlink in the current > runlevel, and avoid starting services if no symlink is present. The symlinks are something internal to sysv-rc. How they are managed is surely your choice. I would like to get an invoke-rc.d that is consistent to /etc/init.d/rc for the current run level. If /etc/init.d/rc doesn't start or stop a service, then invoke-rc.d shouldn't, either. Regards Harri
signature.asc
Description: OpenPGP digital signature
