[Mageia-dev] Any progress on the NFS mount problem? - SOLVED

AL13N alien at rmail.be
Tue Oct 2 12:18:15 CEST 2012

> 'Twas brillig, and Guillaume Rousse at 01/10/12 21:41 did gyre and gimble:
>> With NFSv4, the server remaps its actual filesystem (/home, for
>> instance) in a pseudo-filesystem, which is seen by clients as a root
>> filesystem (/ here). Consider it as some kind of implicit chroot.
>> With earlier implementations, you had to explicitely mark one of the
>> filesystem as the root one, and make sure every other filesystem was
>> reachable from it, using bind mount if needed. You've got some more
>> detailed explanations here:
>> https://wiki.archlinux.org/index.php/NFS
>> According to official NFS wiki, it seems this is no longer needed, and
>> than the implementation handle it automagically for you:
>> http://wiki.linux-nfs.org/wiki/index.php/Nfsv4_configuration#Exporting_directories
>>>  I removed 'fsid=0', restarted, and all is well.  I
>>> can see all my data!
>> Well, you'd better check if you're actually using NFSv4, and not a older
>> version, if you want optimal performances.
> FWIW, I found I needed fsid=0 on my media center's chroot export when
> using an NFS root boot (server + client both mga2).
> Without fsid=0 root was mapped to 65534 which caused numerous issues
> when booting :)
> Never did work out the full reasons, so I probably need to read those
> specs a bit.

FWIW: my mga2 server and mga1 chrooted client don't need fsid=0 . but then
i'm using nfsver=3; don't want to be using nfsv4

More information about the Mageia-dev mailing list