pfs_umount is hung

[Sheriff Andy]

I am having some difficulty with pfs_umount. Is there anything besides a reboot that I can do to clear this?

OS is 11.0.

Points awarded.

thx,
Andy

[A. Clay Stephenson · Accepted solution]

In short, no. You can try killing all the pfs daemons but it ain't gonna work.

[Patrick Wallek]

This is a good reason to NOT use the pfs* commands/daemons.

No there is no way I know of the clear that other than a reboot.

FYI -- The rock ridge extension patches were designed to eliminate the need for the pfs* junk.

For HP-UX 11.0 the patches are:

PHCO_26449
http://www1.itrc.hp.com/service/patch/patchDetail.do?patchid=PHCO_26449&sel={hpux:11.00,}&BC=main|search|

PHKL_26450
http://www1.itrc.hp.com/service/patch/patchDetail.do?patchid=PHKL_26450&sel={hpux:11.00,}&BC=main|search|

PHKL_28060
http://www1.itrc.hp.com/service/patch/patchDetail.do?patchid=PHKL_28060&sel={hpux:11.00,}&BC=main|search|
(Note PHKL_28060 superseded PHKL_26448. PHKL_26448 is listed in the other 2 patches above as being needed for the rock ridge extension to work. If you install PHKL_28060 you will be OK.)

You will have to reboot after these patches are installed.

[James R. Ferguson]

Hi Andy:

Reboot. If you don't terminate the daemons correctly, things go from bad to worse.

You NEED to acquire the three (3) patches for 11.0 that support Rockridge format mounts. While the installation of these patches requires a reboot, you will never again have to resort to this kludge.

Regards!

...JRF...

[Sheriff Andy]

I do have the rockridge patches installed.

I saw this on the usergroup for mount options, you recommend these options?

# /usr/sbin/mount -F cdfs -e -o ro,rr device_file /cdrom

Thx for the quick response.

Andy

[A. Clay Stephenson]

Yep, the -o rr mount option is your boy but, of course, you still must reboot before you can try them new fangled mount options but once you do, throw them pfs critters out of the barn.

[James R. Ferguson]

Hi (again) Andy:

In addition, once you have the necessary patches installed, you can disable the startup of NFS (in 'etc/rc.config.d/nfsconf' if you dont' want/need it. The pfs_mount mechanism leveraged this. This elimiates a few extra daemons if you don't need them and tightens your security too.

Regards!

...JRF...

[Sheriff Andy]

I am running this on another box, and I see everything in uppercase with the mount -o rr option.

Am I missing something?

[Patrick Wallek]

You may or may not have a problem with combining '-F cdfs' with '-o rr'. I have heard of problems, but don't recall experiencing it myself.

If you do receive an error with the command you mention, just remove the '-F cdfs' so that it looks like:

# /usr/sbin/mount -o ro,rr device_file /cdrom

[A. Clay Stephenson]

Ok, umount it and try -o rr,cdcase

[Sheriff Andy]

The files are in lowercase, but now I don't see the root:bin ownership. Is this a concern?

# mount -o rr,cdcase /dev/dsk/c1t4d0 /cdrom
# ll /cdrom
total 268
-r-xr-xr-x 1 -1 -1 534 May 10 2005 autorun.inf
-r-xr-xr-x 1 -1 -1 301 May 10 2005 cdinstall.bat
dr-xr-xr-x 1 -1 -1 2048 May 10 2005 docs
dr-xr-xr-x 1 -1 -1 2048 May 10 2005 j2re_win
-r-xr-xr-x 1 -1 -1 126617 Feb 12 10:43 relnotes.txt
dr-xr-xr-x 1 -1 -1 4096 Oct 5 2006 v61

[A. Clay Stephenson]

Why would you expect to see files which come from a completely foreign machine and proably a foreign OS to match your boxes UID's and GID's -- if the concept even exists on the foreign OS? By convention, -1 indicates unknown and -2 indicates anonymous on most flavors of UNIX. Normally, your installation program is either going to prompt you for a user and group or will assume the effective UID and GID of the installer when copying the files.

[Sheriff Andy]

Thx Clay,

Sounds like I don't really need to concern myself with it.

When I ran pfs_mount earlier, the ownership was root:sys.

Thx again,
Andy

[Peter Nikitka]

Hi,

for mounting DVDs I don't use the option 'rr' at all - only 'cdcase'. If I remember correctly, 'rr' had that effect for such media.
I use 'rr' solely for CDs.

mfg Peter

[Sheriff Andy]

Thank you everyone for your comments.