Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I have a long involved script that is intended to install a Solaris system based on an image. This script was in the "almost working with a couple of minor bugs (like Tee does not exit after pipeline it's on has finished)" state, and now has devolved to "not working", in large part because of the following error.

Everything on the "image" side of the script works just fine. I prepare the system by stopping some services, shutting down and detaching the zones, removing some extraneous files (audit, mail, that kind of stuff), then going to single user mode. There I flar the system, take tarballs of the zones, then reattach the zones and restart the services. That all works fine. Including the zones; they reattach and then boot just fine.

On the target system, however, things get weird. I restore the system from flar, boot it, and then run the script. The script creates volumes on a direct attached array, formats and newfs's the volumes, mounts them, unzips and untars all the tarballs I created, then attachs, boots, reconfigures, and reattaches the zones. This was all working fine, and I cannot for the life of me find anything that I've changed in this part of the script. Now, though, it has started to do the following:

Upon attaching the zone, it complains that it (I'm not using zfs at all).

failed to initialize zfs library

Then, when attempting to login to the zones, I get errors with the /dev system (i.e., /dev/null: cannot open; /tmp/spinning: No such file or directory). Then, after this, when I try to login to the zones to do troubleshooting, I get

zlogin: failed to open /zones/APP1/dev/pts/1: No such device or address

I found some hints that it might be a "device major number" mismatch between the zone and the global zone, but doing ls -lL, the major numbers look the same. The zones show as running with no issues. There are no broken services.

I have been googling for 3 days and I can't even find any hints to follow. Any ideas?

share|improve this question
    
I guess I should also note that as part of the troubleshooting process, I also tried: * manually restoring a zone * restoring a zone and doing an attach -u (that failed when sort bombed out, claiming there wasn't enough something...) * rebooting the zones * rebooting the system * doing a reconfigure reboot on the zone (does that even work?) * deleting the /zones/<ZONENAME>/dev directory and doing a reconfigure reboot * logging in via console (can't get a useful prompt) at all..should I be able to? * check /zones/<ZONENAME>/root/var/sadm/system/logs for anything...there isn't anything –  qkslvrwolf Mar 8 '12 at 14:22
    
Found it. I'm dumb. I had modified a script that was generating the /etc/vfstab with the device numbers for the array volumes. The zones reside on the array volumes. I had added "rw,nodevices" as options for ALL the volumes, including the zone mountpoints. Oops. Zones require devices to work. I removed the nodevices from the options, halted, unmounted, remounted, and restarted the zones, and they work. I'll update the answer when it lets me. –  qkslvrwolf Mar 8 '12 at 20:54
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.