On Sat, 2015-05-09 at 03:32 +0200, Michael Biebl wrote: > On Sun, 03 May 2015 17:23:30 +0100 Ben Hutchings <b...@decadent.org.uk> > wrote: > > Control: reassign -1 clock-setup > > Control: retitle -1 Missing support for systems without battery-backed RTC > > > > On Sun, 2015-05-03 at 17:40 +0200, Geert Stappers wrote: > > > Debian installations on hardware with no (battery-backuped) RTC > > > is likely an installation that hasn't network connection. > > > > What makes you think that? > > > > > So please do not push (too hard) > > > for "you MUST allway known what time it is" > > > > > > Make it possible to do installs on hardware without RTC > > > and no access to a NTP server. > > > > Of course this should still be supported. > > > > > Installing fake-hwclock https://packages.debian.org/stretch/fake-hwclock > > > on the absence of the a RTC > > > > > > Avoiding filesystem checks when no RTC present would also be good. > > > Simular as https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=767040 > > > > Good point. > > > > I'm retitling this because we now have three small related changes > > wanted in the installer: > > > > 1. Install/enable NTP client > > 2. Disable hwclock-save.service > > 3. Disable e2fsck time check > > > > I think these could all be done in clock-setup, so I'm reassigning to > > that. > > FWIW: > http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?h=experimental&id=929bece53261ddd2797d4f3518a05a88704c5b01
So what happens if another ntp daemon is packaged, or they move executables into /usr/bin? > http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?h=experimental&id=5ea1b2bd1c57f9d3095252229c4ba1e50a7248d6 > > We enable timesyncd by default now and have dropped the > hwclock-save.service in systemd for the systemd version targetted at > stretch. > > Is there anything left which needs to be done? Point 3 still needs to be fixed; at least on systems not using systemd-networkd the system clock will still be wrong when fsck runs. Ben. -- Ben Hutchings Theory and practice are closer in theory than in practice. - John Levine, moderator of comp.compilers
signature.asc
Description: This is a digitally signed message part
