On Mon, Mar 13, 2017 at 9:26 PM, Ivan Baldo <iba...@adinet.com.uy> wrote: > Hello, sorry for the delay! > > > El 06/03/17 a las 19:36, Felipe Sateler escribió: >> >> Control: tags -1 moreinfo >> >> On Mon, Mar 6, 2017 at 7:14 PM, Ivan Baldo <iba...@adinet.com.uy> wrote: >>> >>> Package: systemd >>> Version: 232-18 >>> Severity: normal >>> >>> Situation: PXE boot with RO NFS and using OverlayFS for RW /. >> >> Out of curiosity, why do you want read-write root? I figure you could >> just mount /var and /home as rw... > > At the moment I could do with /etc, /var and /home as rw, yes (I need > /etc because of some specific things for some terminals, like printers, X > server woes or something else that needs to be different). > But why not having all rw? Seems simpler and used in live images. > BTW /home is a normal NFS share, not overlayfs.
It might be simpler to apply the usr-merge to the exported filesystem, then only /usr would be mounted from NFS. / could be a real local filesystem. But, I have never done this so take this with a grain of salt. >>> So, I rm'ed /etc/machine-id and /var/lib/dbus/machine-id, thinking that >>> it would be generated at >>> boot and they do, but sys-subsystem-net-devices-multi-user.device >>> timeouts after 1m30s then. >> >> Where does this multi-user.device come from? Maybe the problem is in >> the component responsible of bringing up this net device. > > Maybe, it needs further investigation, but its strange that the behavior > is changed simply by having an empty machine-id instead of just a deleted > one. > Anyway, I grabbed > http://cdimage.debian.org/cdimage/stretch_di_rc2/amd64/iso-cd/debian-stretch-DI-rc2-amd64-netinst.iso > and installed the standard desktop in a VM, deleted the machine-id and > everything was fine, boot was fast, etc. > So, it must be something specific to my environment. Thanks for comfirming this. -- Saludos, Felipe Sateler
