First of all, I can hear you asking, what the #&$*!!?% is a
Netra X1, and why should I care?
Okay, you may not. However, the Netra X1 is a Sun UltraSPARC appliance server, designed for LOM (Lights Out Management). As such, it is bereft of many of the bits we consider 'normal' in a computer, like, say, any kind of removable media drive. This is all well and good, but sometimes Bad Things happen to computers, and you need to fix them. Usually, such things happen at the Worst Possible Time and in the Worst Possible Place, namely, the place where the manual and your own notes aren't. If this happens to you, and you have net access, you might remember that E2 has an answer for you on this one.
The error message in question I got manifested itself during the boot process. The Netra's firmware, during the boot, would announce that the system disk (dad0 in my case) couldn't be read. The reason, according to Sun, is that if the disk label is corrupted, the system firmware isn't smart enough to work around this problem. This is bad enough on the bench, but in an appliance server, there's no CD-ROM drive, and the server may be in a different state altogether!
There are two ways I know of to fix this problem. The first is truly LOM-like, and I've never done it. That is to set up a Solaris Install Server on the same subnet as the Netra (assuming you don't already have one for setting up your Netras in the first place) and then NetBoot the installer from there. The other way, which I have done and will describe in more detail, involves actually touching the machine and adding a CD-ROM temporarily in order to boot from the Solaris 8 distribution/installation CD-ROM.
Adding a CD-ROM
This isn't nearly as hard as it sounds. In bare terms, here are the steps:
- Power off the Netra.
- Unplug the Netra.
- Remove the top cover (note that if you have enough clearance, you need not take it out of the rack to do this).
- Locate the second drive bay. It's on the left side as you face the front. If there's a drive in it, disconnect the drive; the right-hand bay should be the boot drive (assuming default config).
- Take any old ATAPI CD-ROM drive and rest it on top of the hard drive and front panel. Connect the (existant) second IDE cable and power adapter (meant for the second IDE hard drive) to the CD-ROM. Make sure the CD-ROM is set to 'IDE Master'.
- Reconnect the cabling to the Netra (network not necessary; console link is necessary). Using the console connection, you should see the "lom<" prompt indicating that the Netra is on standby.
- Issue the poweron command. This will cause the Netra to boot. It will fail to load the disk as usual, and instead will typically default to trying to boot from the network. When it does, drop back into the lom mode by using the characters #. typed one after the other, on the console.
- Load the Solaris 8 Installer CD (disk 1) into the CD-ROM drive.
- At the "lom>" prompt, type reset -x. The Netra will undergo an external processor reset, and should dump you into a more familiar-looking (if you're a Sun wonk) firmware prompt, which looks like this: ok>
- Type setenv boot-mode cdrom and hit return. You should get the prompt back.
- Type reset -x again. You should drop back to the same prompt, but with the environment change you just made saved.
- Type boot cdrom and hit return. The Netra will undergo boot, but should (instead of trying the disk and failing over to netboot) boot from the Solaris Installer disk in the CD-ROM drive. The first thing the installer will ask you is if you'd like to keep or upgrade your existing install (I can't remember). It doesn't matter; you just want to use the q (quit) option which should drop you to a command prompt. Note: At this point, instructions for CD and Netboot server converge.
- Once you have the command prompt, issue the command format. Don't worry, it won't.
- It should ask which disk you'd like to modify. Select the only hard drive available (which should be your damaged one).
- From the format menu, select Backup. This will locate a backup copy of the disk label on the disk, if it exists. If it finds one, it will ask if you'd like to restore from that backup. Say yes.
- Quit format, preserving changes.
- Reboot. Cross all fingers and toes.
Sun documentation notes that since the corrupted label is stored on track 0 of the disk, it is prudent to
zero out track 0 by adding it to the drive's hardware
defects list, or by simply
reformatting the disk and starting partition 0 at track 1. This is up to you.