Le 22/07/2014 18:24, The Wanderer a écrit : > On 07/22/2014 11:56 AM, Erwan David wrote: > > > On Tue, Jul 22, 2014 at 05:25:22PM CEST, Don Armstrong > > <email@example.com> said: > > >> On Mon, 21 Jul 2014, Erwan David wrote: > >> > >>> I lokked at it. I do not know how to remove this quiet on command > >>> line which seems to have appeared. Did systemd change grub > >>> configuration ? Or did rather change grub semantics ? > >> > >> It's the default in Debian. Edit /etc/default/grub and remove quiet > >> from GRUB_CMDLINE_LINUX_DEFAULT. > > > Fine, Ill test this (and systemd.show_status) at next boot. > > I strongly suspect that this won't produce the desired result. > > The kernel and systemd apparently both react to 'quiet', where sysvinit > does not. This means that you can't pass the 'quiet' option to the > kernel without also passing it to systemd. > > > Under sysvinit, as I understand matters: > > If you use 'quiet' on the kernel command line, you will get init-system > log messages, but not kernel log messages. This was the > Debian-configured default. > > If you leave out 'quiet' on the kernel command line, you will get both > kernel log messages and init-system log messages. > > > Under systemd as I understand it: > > If you use 'quiet' on the kernel command line, you will get neither > kernel log messages nor init-system log messages. > > If you leave out 'quiet' on the kernel command line and don't add > anything else, you will get both kernel log messages and init-system log > messages. > > If you leave out 'quiet' on the kernel command line and add > 'systemd.show_status=false', you will get kernel log messages, but not > init-system log messages. > > As far as I can see, there is no way to get init-system log messages > without also getting kernel log messages - unless either the kernel > starts conditioning its messages on something other than 'quiet' (almost > certainly not happening), or systemd stops reacting to 'quiet' alone. > > Thus, unless I'm missing something, the previous (Debian-configured) > default *can no longer be achieved at all*. That seems like a decidedly > undesirable state of affairs, which is why I hope I am indeed missing > something. > What could work (but I'll have to check) would be having quiet + systemd.show_status=true If it is not achievable, yes it looks to me a problem. All this reminds me the hassle of redirections in *csh... It seems that 20 years later, the lesson of settings which change independant feature is not yet learned...
Description: OpenPGP digital signature