[clug] nfsroot reboot issue

Bob Edwards bob at cs.anu.edu.au
Tue Oct 30 22:40:10 UTC 2018

Hi all,

I have been rolling my own nfsroot image for diskless booting with PXE
for a while, built on Debian using debootstrap. Allows me to trivially
add whatever packages I might need for: rescue image, fresh install etc.

However, recently I have noticed that I can no longer rely on it to
reboot sanely when issuing the "reboot" command over SSH. May be tied
up with now using a read-only NFS share for the root fs, which otherwise
works fine (esp. with a tmpfs mounted /tmp). It is using Debian 9, so
could also be systemd related.

On my test system console, it seems to unmount / before it has finished
running the shutdown scripts and so hangs waiting for NFS to come back.

One server, in particular, is in a lights-out facility and I have no
access to it's console, so being able to reliably reboot is highly
desirable (saves me having to ring someone to push the power button).

Anyone got any good tips/pointers on what I might be doing wrong?

Bob Edwards.

More information about the linux mailing list