ERROR: unable to open mount table /etc/mnttab -Solaris 8 help!

We still have a few servers on Solaris 8. I was testing adding the Time Zone patches for Solaris 8: 109809-04 and 108993-63 on some test systems. Patch 109809-04 installed fine, but 108993-63 says it was missing packages. I was told that all the packages listed under the 108993-63 patch directory needed to be installed before the patch would install correctly. I did pkginfo on all of them:

dssj-stdw-test:hughesm> pkginfo SUNWapppr SUNWapppu SUNWarc SUNWatfsr SUNWatfsu SUNWcarx.u SUNWcarxus SUNWcsl SUNWcslx SUNWcsr SUNWcstl SUNWcslx SUNWcsr SUNWcstl SUNWcstlx SUNWcsu SUNWcsxu SUNWdpl SU

WNdplx SUNWhea SUNWlldap SUNWmdb SUNWmdbx SUNWnisr SUNWnisu SUWNpppd SUWNpppdu

and it said these were missing:

ERROR: information for "SUNWapppr" was not found

ERROR: information for "SUNWapppu" was not found

ERROR: information for "SUNWcarx.u" was not found

ERROR: information for "SUNWcarx.us" was not found

ERROR: information for "SUNWcslx" was not found

ERROR: information for "SUNWcsr" was not found

ERROR: information for "SUNWcstl" was not found

ERROR: information for "SUWNdplx" was not found

ERROR: information for "SUNWlldap" was not found

ERROR: information for "SUWNpppd" was not found

ERROR: information for "SUWNpppdu" was not foun

I installed the first ones, including a prerequisite SUNWpppk up to SUNWcsr which only installed partially.

Now when I try to install any of the other packages I get:

dssj-stdw-test:hughesm> sudo pkgadd -d. SUNWcstl

Processing package instance <SUNWcstl> from </tmp/sunpackages>

pkgadd: ERROR: unable to open mount table /etc/mnttab

No changes were made to the system.

I also cannot remove it:

dssj-stdw-test:hughesm> sudo pkgrm SUNWcsr

The following package is currently installed:

SUNWcsr Core Solaris, (Root)

(sparc) 11.8.0,REV=2000.01.08.18.12

Do you want to remove this package? y

pkgrm: ERROR: unable to open mount table /etc/mnttab

I can't even do a showrev:

dssj-stdw-test:hughesm> showrev -p | grep 108993

showrev: get_env_var(core, SUNW_PATCHID)

The /etc/mnttab file is empty and when I do "df -k" nothing returns. I found some references to this error message on sunsolve but not under the same circumstances.

Help please:

1. How do I resolve this error? Fortunately this is a test box but I have more Solaris 8 servers that are being used.

2. Did I really need to install all these packages just for the Time Zone update? I have more Solaris 8 systems, including a production one that I don't want it to fail on. Is there a way to bypass certain packages on a patchadd is they're not installed and not needed?

[2800 byte] By [hughesm1a] at [2007-11-26 16:08:13]
# 1

Your system is messed up. Reinstall.

If pkginfo can't find SUNWcsr (Core Solaris Root) very little in the way of upgrading or patch installation is going to work right.

If /etc/mnttab is missing or corrupted your /etc partition is probbaly corrupt and you will be lucky to be able to reboot your system. Sorry about that.

wsandersa at 2007-7-8 22:30:34 > top of Java-index,General,Maintenance...
# 2

I think you are right. I've just let this box sit for the last couple of weeks and will probably get around to reinstalling Solaris 8 or installing Solaris 10 . We were unable to reboot off of anything but the CD. Tried to rebuild mnttab but that didn't work either.

What I learned is that on the other production systems we only patched for the existing packages and didn't install any additional packages.

I also suspect this SunBlade 100 had an existing problem because it had crashed a few weeks before this.

Thanks for the response.

hughesm1a at 2007-7-8 22:30:34 > top of Java-index,General,Maintenance...
# 3

On Solaris 8 and higher, /etc/mnttab is a virtual filesystem that points to the kernel. If that mount isn't there, it's just an empty mount point.

You could try:

/sbin/mount -F mntfs mnttab /etc/mnttab

That should recreate the mount if it got unmounted for some reason.

--

Darren

Darren_Dunhama at 2007-7-8 22:30:34 > top of Java-index,General,Maintenance...
# 4

It was at the OK> prompt so I did "boot cdrom" After a few minutes is wants to install the OS (Asked for time zone, etc.) So then I did "boot cdrom -s"Got the following messages - don't know what they mean:

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_ATTACH_REQ(1) errno 14 unix 0

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_ATTACH_REQ(11) errno 14 unix 0

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_BIND_REQ(1) errno 3 unix 0

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_PHYS_ADD_REQ(49) errno 3 unix 0

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_UNBIND_REQ(2) ...

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_ERROR_ACK ...

lr_rput_dlpi(fcip0): DL_ERROR_ACK for DL_DETACH_REQ(12) ...

Using RPT Bootparams for network configuration information.

Skipping interface er10

INIT: SINGLE USER MODE

# /sbin/mount -F mntfs mnttab /etc/mnttab

mount mnttab is not this fstype

(I think I'm missing something here ....probably looking at the CD?)

pwd shows /tmp/root

format shows two disks: 0 c0t0d0 and 1 c0t2d0. I think we determined that c0t2d0 was actually the boot disk and c0t0d0 was the mirror. Seemed backwards.

hughesm1a at 2007-7-8 22:30:34 > top of Java-index,General,Maintenance...