In rc2 (and previous betas AFAIK - didn't have the time to have a closer look at this), there are error messages when /usr is (read-only) on an NFS server. On shutdown, the following messages are printed (including the messages before and after the errors): Stopping crond: [ OK ] Unmounting NFS filesystems: umount2: Invalid argument umount: /usr not mounted [ FAILED ] /etc/rc.d/rc: line 125: 1616 Killed $i stop Saving random seed: [ OK ] ... Shutting down system logger: [ OK ] Unmounting NFS filesystems: [ OK ] /etc/rc.d/rc: line 125: 1748 Killed $i stop Flushing all chains: [ OK ] ... Resetting built-in chains to the default ACCEPT policy: [ OK ] Starting killall: /etc/rc.d/rc: line 4: 1876 Killed initlog $INITLOG_ARGS -c "$*" [ FAILED ] Sending all processes the TERM signal ... [ OK ] ... Unmounting file systems [ OK ] umount says, /usr is no longer mounted. I checked this, umount is lying here ... :) The first "[process] killed" comes from /bin/bash /etc/rc6.d/K75netfs stop The second "[process] killed" is /bin/bash /etc/rc6.d/K90network setop The third one is initlog -q - /etc/rc6.d/S00killall start I did try a fuser -m -v /usr right after these messages. The first fuser (netfs) says, that K75netfs is using /usr (because fuser is running from this script???). The second one (network) says, that K90network and netfs are still using /usr. The third one (in killall) says that initlog and S00killall are using /usr. I did a cat /proc/mount before and after each of these messages. /usr keeps being mounted.
The problem still exists in the final release.
The problem still exists in Hampton beta3.
Closing bugs on older, no longer supported, releases. Apologies for any lack of response. If this persists on a current release, such as Fedora Core 4, please open a new bug.