Random stuff from my week

This has been quite a busy week, both at work and at home. Having to children at home keeps you busy at the best of times, starting to prepare for the new baby (due date October 21 folks!) is keeping us in a permanent state of panic.

Work has been busier than usual. Last week was spent on a fast-track Windows Server 2003 Active Directory Planning, Implementing and Maintaining course (Microsoft 2279B and 2282A). Since I've spent the last year or so battling with our current AD setup most of this was pretty familiar, but it is nice to get an insight into the less well used aspects of AD, and also to see the new features for AD 2003. Now all that remains it to persuade my boss to buy the licenses to upgrade our AD controllers to Windows Server 2003.

This week has been varied at work (to say the least). We had a major crash on our main editorial server (running on Solaris) that took a little while to sort out. The problem looks like it was either caused by the network card or by some corruption in the /devs directory. Either way the outcome was the same - neither of the network cards would start on boot-up. After removing and re-seating the gigabit card I was able to start it manually.

After trying quite a few alternatives (checking the network and so forth) we did a "boot -r" to recreate the device nodes, followed by another reboot which got everything up and running again.

This was followed by spending some time upgrading our backup Editorial server to give it enough disk space to adequately hold a copy of our editorial data. Finding out how to get the extra disks added to the A1000 controller was fun, to say the least. To cut that story short you need to be running Array Manager 6.22.1 to get it to run on Solaris 8. A word of warning - don't be tempted to install an earlier version. I did, and had one hell of a time getting access back to the existing raid arrays.

Luckily for me Sun still provides downloads of Array Manager 6.22.1 as well as some decent documentation on installing and running it.

One problem I did encounter is that once we had installed the new disks, created a new LUN and a raid 5 stipe set (with one disk left over for a hot spare) we couldn't create a file system on the disks. In fact we kept getting the error that the disk geometry was unrecognised.

Running format against the new disk set provided the fix for this - the labels on the newly added disks were incorrect. Once they had been re-labeled by format it was a simple task to create the new file system and migrate over the old data from the old disk to the new one.

To top off the week was my yearly assessment. A couple of hours of being told how wonderful I am by my manager is great. Now about that pay rise. . . . . .

. . . . .


To finish off with I've just got round to upgrading the hard disk on my current test laptop. It has a 15gb drive, but I wanted to use the 40gb one from my previous tester. Swapping the drives is the work of a few minutes. How about re-installing the OS?

The original machine dual booted between Windows XP and FreeSpire, the current one was just using Ubuntu. Booting up the new laptop was great. FreeSpire autodetected all the new hardware, went straight into X, started the onboard wireless and connected to my network. Amount of reconfiguring to do - none. Everything works perfectly and at the higher resolution that the new laptop supports.

Now for Windows XP. The story here is not so good. Windows "Blue Screens" about two seconds into the boot process. It won't go into safe mode either.

I'll have to decide whether XP goes back on when I re-format to put back on Ubuntu.

Fun with a Sun

One of the joys of working in I.T. is the sheer variety of tasks that can await you.

Today I've had the pleasure (ahem) of dealing with a disk upgrade on a Sun A1000 raid controller, connected to a Sun Ultra 60 running Solaris 8.

On paper the job is straight-forward. Simply add the new disks into the array, use them as a raid 5 set, configure a new partition in the new container and adjust the vfstab to mount the new partition at the correct point on the system.

Easy, no?

Stage 1 was started last night. To check how our raid was currently set up we went to check the Raid Array Manager software that comes with the A1000. Not installed. How about disksuite them? Nope. Not installed either.

We couldn't locate the Solaris 8 install CDs but found the Raid Array Manager install CDs so we installed that, and as per the instructions halted and re-booted using "boot -r". And heres where the fun begins.

Now it couldn't see the existing partitions on the Raid, even though the Raid Array Manger showed that the Raid was connected and that the individual drives were healthy.

If you ever find yourself in the situation where you need to see what SCSI devices are attached to your Sun box then this might help.

After halting your box (so you are left at the "ok" prompt) do the following:

setenv auto-boot? false
reset-all
probe-scsi-all
setenv auto-boot? true
boot

Depending on the circumstances you may need to do "boot -r" instead of just "boot". The difference is that "boot -r" will detect and re-create device nodes as needed.

After a bit of investigation (thank you Google) I removed the Raid Array Manager software, at which point "boot -r" rediscovered the existing partitons. For some reason though it has re-named the device nodes for the RAID disks so I still had to edit the vfstab (after creating a backup of the old one just-in-case) before the server would boot again correctly.

So the alternative is to locate the Solaris 8 media and install DiskManager.

I can see why my colleague had trouble locating the disks. In the end I found them in an un-marked cardboard box, on top of a disused filing cabinet, right at the back of our software store-room.

Disksuite is now installed and the partitions are still visible, so we are now ready to try putting in those new disks. . . . .