When the nfs is started, the init.d script mounts the nfsd on /proc/fs/nfs type nfsd (rw,noexec,nosuid,nodev) However, when the /etc/init.d/nfs is stopped, one would expect that everything relating to NFS service provision would be stopped as well, atleast, what was caused by the actions of the init script. Unfortunately, the proc is left mounted, thus preventing the nfsd module from unloading. and you know how tiresome it could be sometimes to find out what keeps the module busy. Reproducible: Always Steps to Reproduce: 1. /etc/init.d/nfs start 2. /etc/init.d/nfs stop 3. lsmod | grep nfsd nfsd 265256 1 auth_rpcgss 45408 1 nfsd exportfs 5376 1 nfsd nfs_acl 3840 2 nfsd,nfs lockd 69296 2 nfsd,nfs sunrpc 184392 7 nfsd,auth_rpcgss,nfs,nfs_acl,lockd 4. mount | grep nfsd nfsd on /proc/fs/nfs type nfsd (rw,noexec,nosuid,nodev) Expected Results: include umount /proc/fs/nfs in the stop part of the script
we dont do pseudo filesystem or module cleanup with init scripts ... no real way of knowing what the user's intentions are with either.
