There is no reason to try to remount a nfs-mounted root filesystem readonly, since fsck can't be run on a nfs filesystem, and it just makes an ugly failure message appear in the startup of a gentoo system with a read-only nfs-mounted root filesystem. Reproducible: Always Steps to Reproduce:
Created attachment 62965 [details, diff] patch for checkroot to make it not attempt to remount a network-mounted root filesystem
Hmm, I wonder in any case if we should not make it more generic (check if its already ro, and do not remount if so), as currently it will remount it anyhow, upping the mount count, and causing journaled filesystems to be checked in half the reboot counts that they should have been checked ....
Created attachment 62968 [details, diff] checkroot.patch i think we should do both ... how does this look to everyone ?
That looks pretty good to me, I thought about doing something like that, but decided to just check for a network FS in my patch, assuming there was some reason in particular the script wasn't checking for already ro filesystems, other than not being implemented yet.
Yup, ok here as well. Just remember to fix that inverse if ! touch ... logic.
fixed touch logic and committed