I wrote:
> Server's exports:
> /usr/local -ro
> Server's fstab:
> /dev/sd7g /usr/local 4.2 rw 14 3
> Client's fstab's:
> <Server>:/usr/local/long /usr/local/long nfs ro 0 0
> <Server>:/usr/local/bdd /usr/local/bdd nfs ro 0 0
> Now all of a sudden (and several Days after the Mount, but immediately
> after a Crash&Reboot of the Server), my TCP Wrappers send me Alert
> Mails because the Clients try to contact the Server's rpc.rquotad.
> Now, is contacting the quotad responsible *for a non-quota'd read-only
> NFS-mounted FS* what one should expect, or is something fishy going on?
Turns out that I've got to learn how to RTFM. :-} As Dave Mitchell
<D.Mitchell@dcs.shef.ac.uk> pointed out to me, the (no)quota Options
are different for local Disks and NFS Mounts; NFS Mounts are "quota"
(which doesn't exist there in the first Place) by Default, while local
Disks are "noquota" by Default, which I mistakenly assumed for NFS as
well. I added the "noquota" Option to the untrusted Clients' fstabs,
and hey presto!
I meanwhile realized, too, that TCP Wrappers for rpc.rquotad are not
airtight since rpc.rquotad stays running and accepts further Connections.
This explains why the Alarms went off after rebooting the Server, instead
of when the Clients mounted Stuff the first Time.
Eric William Burger <ericb@telecnnct.com> mentioned that replacing
/usr/ucb/quota with a Link to /bin/true works as well, but might not be
appropriate for all Installations. :-}
Thanks again,
J. Bern
-- __/\___________________________________________________________________________ / \ Mail: bern@Uni-Trier.DE (7Bit); bern@TI.Uni-Trier.DE (OW Mails OK); /\ / J. \ My Homepage / \ \Bern/ Finger to this Site's broken, Lotsa other Things too, use Mail+WWW! \ / \ / P. O. Box 1203, 54202 Trier, Germany _____ Ham Call Sign: DD0KZ \/ __\/_________________________________________/ EOF \___________________________
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:14 CDT