Hello, Am 12.02.2014 01:06, schrieb [email protected]: > Dear Maintainer, > > even though I have RAMTMP=yes in /etc/default/rcS, recently (~2w ago?) > installed systemd will not mount /tmp as tmpfs. > > Of course, /etc/default/rcS is not a configuration file of systemd so I > wouldn't consider a bug that it doesn't look there. However, I believe > that to provide smooth upgrade experience, basic system configuration > should be preserved during a switch to systemd.
You are right, /etc/default/rcS is not a systemd configuration file and
we deliberately don't read that file.
We do ship a tmp.mount unit though (disabled by default), which you can
easily enable via "systemctl enable tmp.mount"
That said, it seems those settings are stored in /etc/default/tmpfs
nowadays, not /etc/default/rcS
> Therefore, I think systemd install scripts should configure systemd
> according to settings specified in /etc/default/rcS, or at least query
> the user in case there'd be a good reason to switch away from /tmp as
> tmpfs during an upgrade to systemd.
We could do a one-time migration of that setting in postinst, similar to
how we handle the UTC or TMPTIME configuration. This is a simple patch,
something like that
# Do a on-time migration of the RAMTMP setting
if dpkg --compare-versions "$2" lt "204-8"; then
if [ -f /etc/default/rcS ]; then
. /etc/default/rcS
fi
if [ -f /etc/default/tmpfs ]; then
. /etc/default/tmpfs
fi
if [ "$RAMTMP" = "yes" ]; then
systemctl enable tmp.mount || true
fi
fi
I deliberately use systemctl here, and not _systemctl(). "systemctl
enable" should work even if systemd is not the active PID 1
I'm not sure if we should read both /etc/default/rcS and
/etc/default/tmpfs or if the latter only would be sufficient.
Tollef, Michael, thoughts?
--
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
signature.asc
Description: OpenPGP digital signature
