Our Next Upgrade Will Be A Priest

Preferably and old one who only speaks Italian and has lots of first-hand experience with exorcisms. Better yet if he wears a funny hat.

On top of all the other bad luck and weirdness this year, yesterday started out by refitting a whole network closet. We sprung a leak overnight and lost two Cisco switches, a midspan PoE injector, and the UPS that kept them lit. So the whole science floor was out of commission, wired and wireless, until we dried everything out and replaced the electronics. Fortunately, we had lots of canned air and a hair dryer from the time an air conditioner barfed all over our servers, and we also had two spare switches ready to be swapped in with minimal reconfiguring. Our spare UPS, however, was dead dead dead, so we ended up putting everything on a power strip. Not perfect, but everything was operational again by 9:30 am.

This kind of crap has been par for the course this year. I figure that someone must’ve really pissed off the gods this summer. If nothing else, though, at least we’ve learned how to work fast under pressure.

6 Comments Add Yours ↓

  1. Tom McNicholas #

    It happens.

    I had two switches just up and stop responding on me earlier this year. No rhyme or reason. I think they both had the same batch of flashram in them that went bad, as their serial numbers were very close.

    In other news I have a 10.3 mac absolutely, completely, refusing to run the colorsync utility.

    I’ve tried everything under the sun. Just. Won’t. Do it.

  2. 2

    Switches do tend to die in batches, don’t they? We haven’t had any go out lately (except for the ones that got doused), but I was holding onto a couple spares to replace ones with noisy fans. Now we’re out of our spares, so we’ll have to get some more to have on hand before the noisy ones give up. Which will probably be tomorrow.

    As for ColorSync, I’ve got to ask: you did try cutting a new account and running it there, right?

  3. Tom McNicholas #

    I’ll give that a shot, though it does it right now under the original “admin” account the machine was installed with, and also under the user’s account who also is an admin.

    This is a known bug for 10.4, but not 10.3.

    Further driving me nuts, is that this machine was prepped by carbon copy cloner from my “source” G4 that is pretty clean. That machine doesn’t have a problem, nor do a few others that have been cloned off the same source. Weird eh? I’m thinking an Adobe app or system update is the culprit. I may reinstall the 10.3.9 combo update and see what happens.


  4. 4

    Your comment reminds me of some similar problems that we had a few years ago, when we were still running 10.3. Sometimes, on some computers, certain programs just flat out refused to run. They were all imaged the same, so the problem wasn’t the source, and every time we re-imaged a problem computer, the nonfunctional programs remained nonfunctional. In the end, I decided that it was due to bad blocks on the hard drive, and the problem programs were always getting written to the same blocks. Unfortunately, I don’t know if I ever verified that hunch or what we did to resolve the problem.

  5. Tom McNicholas #

    You might be on to something there. I have seen Windows boxes throw fit after fit, ghost image after ghost image no rhyme over reason.


    You zero fill the hard drive, and BAM, fixed. Sometimes using a quick erase on a Western digital to kill the first and last million blocks does the trick. I suppose there is a Mac equivalent of this somewhere.

  6. 6

    Zero Out Data:
    Disk Utilities > Erase > Security Options…

    No idea how to do the Western Digital thing that you did.

Your Comment