NFS woes . . .

NFS woes . . .

Post by Alan Whine » Sun, 28 Jan 1996 04:00:00



: In days of yore (Fri, 12 Jan 1996 05:54:00 GMT)

: :Hello All,

: :I have a Caldera/RedHat 2.0 system that will not allow NFS mounted
: :directories to be written to. I've tried mounting from another
: :Caldera/RedHat system, as well as a FreeBSD 2.0 system and get the
: :same results.

By default, NFS will not allow root (UID 0) to write any mounted file systems
and will only allow users with the same UID as the owner on the mounted
drive to write to mounted systems -- the point being that if your UIDs don't
match for a given login on either system, you need to make them match.

Quick/dirty/risky UID change:

Look at the passwd line on system 1:

joe:Hhs8HhJKlkjl:501:100:Joe User:/home/joe:/bin/csh

and the line on system 2:

joe:Hhs8HhJKlkjl:4099:100:Joe User:/home/55/joe:/bin/csh

and then, on the machine you want to change, (just for fun LOG IN as root,
don't "su"if you are joe...), change the UID in the /etc/passwd file, go to /home and do a
chown -R joe joe.

If there are things in joe's tree that shouldn't be joe's, you should probably
unmount them or do the chown a directory at a time.

: :I believe that my exports file is correct, since I can mount the remote
: :volume. Even when I explicitly add "rw" to the mount command, if I go
: :to write to the directory (as root or as a user), it denies permission
: :to write the file.

: :This is particularly troublesome since the directory mounted is the
: :/home directory for users! Can anyone shed light on what I might be
: :doing wrong? Any help will be appreciated.

:       What are the permission of the home directory *before*
:       anything is mounted on it (and it is exported)?

: :Thanks
: :Frank Kelly

: --
:          />                   JaDe     |     Star                 <\
:         /<                         \|/                          >\
:  *[/////|:::====================- --*-- -=====================:::|\\\\\]*
:         \<                         /|\                          >/

 
 
 

1. NFS woes

I have an RS/6000 Model 320 running "Golden" code.  I tried to nfs mount
a filesystem from a Gould/Encore 9000 running UTX/32 2.0 (BSD 4.3 derivative).
The system seemed to mount ok but I was getting error messages on the
console of my 320:  clnttcp_create: not found
                    rpc.statd: cannot talk to statd at remote_host_name
Performance on the 320 was also degraded severely.

I tried to unmount the remote filesystem, but the filesystem was busy
and I couldn't kill the processes using it.  I rebooted the 320 then
tried to "Stop using the mounted file system".  SMIT could not find it.
I also looked in /etc/fstab.  No entries for the remote machine were
there either, but I was still getting the error messages.  I tried
stopping NFS.  This "failed" with "Subsystem inactive... time out;"
although, the messages did stop.  When I restarted the NFS daemons the
messages and slow response returned.

Now with the Gould down, I still get nfs error messages.  How do I get
smit to recognize that I no longer want to nfs mount the remote filesystem?
Where is it keeping the information on the remote machine?  It's not
in /etc/fstab, and I grepped through all the rc files I could find with
no success.

Any ideas?

Thanks,

Tom Kile

2. problem with installing solaris x86 2.8

3. Time problems

4. NFS woes: tcsh's internal ls fails on Solaris 2.6 client / IRIX 6.x server

5. Two Ethernet cards in RH 6.0 - Only one will work

6. NFS woes . . .

7. Getting Login Id from a program

8. NFS Woes - summary

9. more NFS woes

10. secure nfs woes

11. NFS woes

12. Slow NFS woes