Digital Diatribe

What kind of server comes with make but not gcc or any tools to compile stuff from source?

I run make, and get complaints about gcc being missing, so I apt-get install gcc and then get errors about stdio.h being missing, which is never a good thing when you’re compiling C. So I google for the package I need and come across a suggestion to download build-essential.

If build-essential is so fucking essential why the fuck don’t they include it in the installation? Do they think you’re never going to install anything on your spanky new ubuntu server?

It’s just re-enforcing my views on ubuntu not being all that great for getting shit done out of the box in a server environment.

EDIT: I just realised this only happens on the 64 bit version of ubuntu server. I went to do an apt-get install of build essentail on the i386 version and it was already there.
I guess it’s because there are a few versions you can run on the 64bit machines, which makes it tricky.
Ubuntu is not that bad really :hugs:

This is the first in what will hopefully be a new trend with me writing about my new job and all the exciting things I get to destroy^^^^^^^play with

The PERC raid cards and the powervault 114T both ship with the same default SCSI ID of 6.
Now normally, this wouldn’t be an issue if you ordered any of these devices separately, and tried to plug them into a poweredge 2850.
This is because the poewedge comes with it’s onboard SCSI controller set to an ID of 7.
If however, you would rather use the dual channel PERC card because it’s faster than the onboard one, you’ll have to remember to change the SCSI ID of the tape drive.
Which is where we start today’s lesson…

Not soon after I’d powered on the tape drive I heard a beeping sound coming from somewhere.
I’d been trying to run some commands on the server and had been encountering I/O errors so I thought I’d better go and have a look.
I very soon realised that the beeping sound was emanating from the new RAID controller card.
Looking at the box, I saw some brightly flashing orange lights and after looking up the audible error code in the manual found out that two of the disks had failed.
I thought this very odd and somewhat coincidental, but it still took me a while to work out exactly what had gone on.

I tapped the keyboard attached to the server and saw that one of the drives had failed and the system was now running in read-only mode. Which basically meant I couldn’t even issue a shutdown command So I powered it off and went into the PERC bios to have a look.
It was reporting that both the hot spare and the first disk in the system array had failed.
This is where my memory of events gets a little fuzzy.
At some point I noticed that instead of PROC, ID 6 was reading TAPE. This is what tipped me off that something had gone wrong.

I tried booting on the remaining drive in the array, but this wouldn’t work. It looked like a few of the drives got scrambled when plugging in the tape drive, as no matter how many times I swapped the hot spare and the system drives they would always show up as the same drive in the same channel, which seemed odd, as the Dell tech I was now on the phone with assured me the controller could keep track if you moved drives from one drive bay to another.
What we eventually did was force one of the failed drives back online, and turn the one remaining ready drive into a hot spare.
We then powered off the powervault and made sure the system would boot using only the one drive, which worked fine.
This allowed me to boot the system in a degraded state and simply wait for the array to rebuild.

After about two hours I turned the powervault on only to find that the first 5 disks in the array were now all reporting as failed and there was an NVRAM mismatch error.
This was solved by simply saving the current disk configuration, but the drives were still all reporting as failed.
I knew none of them had in fact failed, so I decided to force them all back online and reboot.

I then changed the SCSI ID of the tape drive to 7, plugged it back in, turned everything on and magically I can access all my drives again.

This whole process took the better part of 7 hours.

So remember kiddies, always check the SCSI ID of a device for conflicts before plugging it in.
I have recommended to Dell that they not ship the PERC cards and the tape drives with the same default SCSI ID, although I’ll happily admit I should have checked to see what the PERC default was instead of assuming it was 7 like the onboard LSI card.
I never thought that a duplicate SCSI ID could cause such a catastrophic failure though. Live and learn, as they say.

So, I got sick of waiting for this cheque from the Strathfield CFO that Brook said was coming and finally decided to do something about it.
So I got in contact with the community law service at uni and they finally got in contact with someone at Strathfield who wasn’t an outright bullshit artist. This person offered to immediately refund the cost of the ipod, which was nice, but I asked for some form of compensation for being stuffed around for 9 months.
At first, they offered me a 10% discount on any future purchases.
So, I’d give them a stack of cash, and then they’d give one 10th of it back to me, and that’s meant to be a form of compensation. I think not!
So after sending the new general manager (did I mention Brook left without telling anyone anything about my case?) an email explaining just how much his company has put me through, we managed to come to an arrangement.
I asked for a Sony Ericsson W810i, which retails for anywhere up $799. I got this instead of the $499 refund. Some people tell me I should have asked for more, and maybe I should have, as I know the phone would probably have cost them less than the refund. But it would have cost me more than $499 to pick one up, so I see as a good deal. If only I could have got a case and car charger with it…

More on the phone later, as it’s pretty awesome. I’m just glad this saga has finally drawn to an end.