NFS mount failure

[Ryan Hurst]

I get an error when this command runs during a restore:
Note: Retrying: "/sbin/fs/nfs/mount -oro 150.1.10.90:/A1/swill /tmp.ign_configure/archive_nfs"
Permission denied

Where the 150...90 is my ignite server address, swill is the hostname of the client (and it's archive directory on the server). The A1 directory corresponds to a disk on a jamaica drive tower. I do not have this issue if the archive is stored on the internal disk. I made sure that the A1 directory and the swill directory both have full (777) permissions, there is an entry for the directory in the /etc/exports file and has been re-exported. I don't know what else to look at to make this thing happy.

[Stefan Farrelly]

Check the permissions on the underlying A1 directory when it is unmounted. The permissions you see when it is mounted may NOT be the same as those when it is unmounted. It may be set to non-777 when you unmount it. If so then change it and mount it again.

[Ramkumar Devanathan]

hi,

check this -
# showmount -e 150.1.10.90

this should tell you if there is an exported share on 15..90 over which you have access rights.

- ramd.

[Ryan Hurst]

Ramkumar:
I did that command. It showed /A1/swill, as well as some other archive dirs.

Stephan:
I didn't use mount, I configured the disks in SAM, and therefore they don't show up in mnttab and I can't umount them. Maybe I should undo them in SAM and use mount, since they are technically hotswappable/removeable?

[Bill McNAMARA_1]

check you /etc/exports on your ignite server.

You need the -anon=2 option.

for the archive directories/directory

Later,
Bill

[Ryan Hurst]

Bill:
Thanks, I did that already though.

[Bill McNAMARA_1]

ps -ef | grep -i nfs

on the ignite server.
Make sure it's running.

--

$ cat /etc/exports
#
#
/A1 -anon=2

$ exportfs -av

--
in inetd.conf should have:
tftp dgram udp wait root /usr/lbin/tftpd tftpd /opt/ignite /var/opt/ignite
.
.
.
instl_boots dgram udp wait root /opt/ignite/lbin/instl_bootd instl_bootd
~
~

nfsstat for info on NFSd status.

Later,
Bill.

[Ryan Hurst]

The output from ps -ef | grep -i nfs:

root 704 0 0 Apr 11 ? 0:00 nfskd
root 6807 6039 0 08:41:47 pts/ta 0:00 grep -i nfs
root 1540 1 0 Apr 11 ? 0:02 /usr/sbin/nfsd 4
root 1544 1540 0 Apr 11 ? 0:02 /usr/sbin/nfsd 4
root 1546 1540 0 Apr 11 ? 0:02 /usr/sbin/nfsd 4
root 1547 1540 0 Apr 11 ? 0:02 /usr/sbin/nfsd 4


cat /etc/exports shows this:

/var/opt/ignite/clients -anon=2
/var/opt/ignite/recovery/archives/jen -anon=2,access=jen
/var/opt/ignite/recovery/archives/igtest -anon=2,access=igtest
/storage/archives/jen -anon=2,access=jen
/A1/swill -anon=2,access=swill

These have all been exported using exportfs -av.
Does A1 also have to be entered in /etc/exports as well as the actual subdirectory used to store the archive?


The relevant lines from /etc/inetd.conf, exactly as they apear in that file (except the elipses):

tftp dgram udp wait root /usr/lbin/tftpd tftpd /opt/ignite /var/opt/ignite
...
...
...
instl_boots dgram udp wait root /opt/ignite/lbin/instl_bootd instl_bootd

[Ryan Hurst]

Sorry, that last part didn't come out right. The /opt/ignite\ and /va/opt/ignite show up on seperate lines below the tftp line. I'm resonably sure that this isn't a word wrap issue with vi, as I expanded the window to max width to make sure. For whatever reason when I ftp'd, copied, and pasted the lines onto my reply window they all got put on the same line.

Is thsi normal behavior, or could it be causing the problem?

[Michael Steele_2]

Look for control chars:

cat /etc/inetd.conf | col -b > /tmp/file_inetd

Note the byte count and diff the two.

[Steven E. Protter · Accepted solution]

You need to make the above recommended -anon changes and re-run exportfs -a on the server.

You might want to make sure networking is right because network inconsistency commonly causes this error message, which I consider an old friend.

A rule: You can not boot an Ignite client off the Ignite server except using the built in lan card, lan0.

After the boot, you can do the ignite transfer through any card you want.

instl_adm -d on the server to display the information.

For this example, the nfs mount and ignite transfer is going to be lan2, adapt for your system.

lanadmin -x 2

Check the results. If they are not the maximum or expected values you have two choices:

1) lanadmin -X 2 100FD if 100 BaseT full duplex is your destired results.

2) Hard code desired settings in /etc/rc.config.d/hpbtlanconf and boot your box.

3) If your switch was built by Cisco, turn off auto-negotiate and have the Cisco admin hard code the desired settings on the port for the Ignite server and the Ignite client.

I have had and overcome your issue and would be happy to help. I'm about to hit the highway to get home for Passover but will check this case when I get there.

SEP

[Ryan Hurst]

Michael:
Here's the byte counts. In vi at least, they are visually identical. I don't totally understand what that command you gave me does...

# ll /etc/inetd.conf
-rw-r--r-- 1 root sys 4395 Apr 1 09:50 /etc/inetd.conf
# ll /tmp/file_inetd
-rw-rw-rw- 1 root sys 4073 Apr 14 13:50 /tmp/file_inetd
#

I have attached the output from running diff on the two files. It's about as long as the files themselves, so there may be a sneaky newline early in one of the files or something making them appear to be totally different files.

Steven:
I too am going to have to head home in a moment. I haven't had a chance to digest and try out your response yet due to other job duties today, but I will return to it tomorrow morning, and I will no doubt have more info and questions at that time. Thank you very much all of you.

[Ryan Hurst]

Oops, forgot the attachment...

[Ryan Hurst]

I won't be able to get back to this thread for a day or two, my project list has been once again "re-prioritized". I will post again when I am able to try out the last set of suggestions...
Thanks again everyone.

[Ryan Hurst]

Thanks! Apparently I needed to export the /A1 directory as well. The recovery ran without a hitch.