i have searched a (too) long time to fix a nfs mount problem. But there was no problem with NFS server/client. Only the “df” command shows the wrong (not expected) output.
Only with the “-a” option shows all nfs mounts from a server.
[CODE]dmz2:/ # df -t nfs
Filesystem 1K-blocks Used Available Use% Mounted on
10.1.1.108:/DATA/tmp_data/test1 51606528 20273664 28711424 42% /mnt/1
dmz2:/ # df -a -t nfs
Filesystem 1K-blocks Used Available Use% Mounted on
10.1.1.108:/DATA/tmp_data/test1 51606528 20273664 28711424 42% /mnt/1
[COLOR="#FF0000"]10.1.1.108:/DATA/tmp_data/test2 - - - - /mnt/2[/COLOR][/CODE]
This does not happen in opensuse 13.2! SLES10:
df --version
df (GNU coreutils) 8.22
Opensuse 13.2:
df --version
df (GNU coreutils) 8.23
I just want to tell this, so that others do not waste time to search for NFS problems
As of September 25th 2015, it looks like a there is a coreutils package update in SLES 12 maintenance which improves this situation, though I don’t know the full details of how it is designed to work now.
The issue described in this thread was not necessarily a bug, it was more like a side effect of certain policy decisions. (But I’m not saying whether the decisions were correct or not.) There have been some concerns and arguments about whether to list all separate NFS mounts, if/when they actually point to the same file system on the server side. Df is supposed to report info on unique filesystems, not necessarily on unique mount points. Listing things in a duplicate fashion can lead to inaccurate totals. But for now, I’m throwing out all philosophical arguments and just saying that an update is available which represents SUSE’s current intent of how this should work.