I don’t need this.

Last night was the third Sunday in four where my UPS has woken me up with beeping. Each time, it seems to suddenly decide that while the load is unchanged, and the charge percentage is unchanged, the projected lifetime in the event of a power loss has suddenly changed from 100 minutes to 0 minutes. This is more than likely an indication that the batteries are failing, and it’s time to replace them. The UPS does some sort of self test once a week, and evidently this one does these tests at 1:30am on Sundays. There doesn’t seem to be any way to turn off these tests or reschedule them.

I spent a hell of a lot on this UPS, wanting one with a lot of capacity and which had replaceable batteries, because my previous one hadn’t let me know that the batteries were getting old until our power went out one day and the charge hadn’t lasted long enough to get down to the computer room to shut down my servers. I can’t remember what I paid for it, but the current equivalent model retails for around $600, so it was probably up in that range. Replacement batteries seem to be around $50 with $20 shipping, and plus then I’d have three small lead acid batteries to dispose of somehow.

On the other hand, I’ve migrated a lot of the things I wanted a Linux server for from my home to my colo box. I’m starting to question if I even need a 24×7 server in the home. Maybe rather than spending all that money on a UPS, I should just move the last remaining things (the mail server and the personal web pages) to the colo box and shut down my home server. It’s a shame to trash a $600 UPS for want for $60 in batteries, but maybe I can eBay it.

I did not need that.

Last night, my UPS started beeping in the middle of the night. This happened once before recently, and that time I just pushed the button on the front to see if it would reset the problem, but it turned the power completely off. That time, after my linux box booted, the two external USB backup drives came up really slow, which caused the hourly rsync backups to take more than an hour, which caused all sorts of hilarity. So this time I decided to shut everything down gracefully before I reset. And yet, this morning I got up to find that four hourly backups are still running.

I killed all the backups, unmounted the usb drives, ran fsck (which didn’t do anything because it said they were clean), powered them off, powered them on, made sure it said they were “high speed” rather than “full speed”, and mounted them. And yet when I did an ls on each one, it hung for over a minute, and then had a message in the log about resetting the USB controller, and then it was fine. I’ve started an hourly backup, and it’s taken 15 minutes or more already and it’s still on the first drive. That’s not good. I wish I knew what was going wrong there.

My first order of business is probably to order new batteries for the UPS. Each time it starts beeping in the middle of the night, munin tells me that the “charge percent” has stayed at 100%, but the runtime in minutes has dropped to zero. Very odd.

It’s definitely the hardware!

I booted my laptop with the install DVD, started up the Airport, and fired up Network Utility. I used Network Utility to continuously ping my Linux box, and put it down. Then Vicki and I went out to dinner. When I came back, it was still happily pinging, but as soon as I picked it up it the Airport died. I guess that proves once and for all that it’s not Parallels or something else in my installation.

You know, it would be a real shame if it was so broken that they had to give me one of the new glass screen MBPs, wouldn’t it?

More fascinating facts about my Airport problems

It seems to happen more when I’m holding the laptop or moving it around, which would lead me to think that it could be flexing and dislodging something lose. It happened once today while I was at work and it was on my desk, but my home office is so cold I wouldn’t rule out thermal flexing. It failed tonight while I was standing there holding it in the Apple Store waiting for it to finish booting, it failed again while I was carrying it to the dinner table, and it failed just now when I picked up my computer from the floor.

It happened when I was logged on as Guest, which was something the Apple Store Genius suggested I try.

Whenever I reboot it because of this, Spotlight is busy indexing and can’t be used for a few minutes after it comes up. This is slightly annoying, because I use Spotlight more than the Dock to start up programs. But more than annoying, it’s worrisome because I don’t remember that happening before.

Tomorrow I’m going to try booting it from the install disk and seeing if it fails while it’s running off the DVD.

Sitting in the Apple Store…

I’m sitting at the Genius Bar waiting for the Genius to come back from the back room where he was probably googling my symptoms.

And now he’s back with some strange theories about a magic keyboard shortcut that disabled my Airport. And now he’s installing some third party software called OnyX. He keeps saying “well it’s still connected” and I keep reminding him that yesterday it failed three times in 5 hours (ie about once every 1.66 hours) and we’ve only been looking at it for half and hour, so that’s no indication of anything.

Oh well, next stop I’ll have to try something more drastic like ‘archive and reinstall’.

Update: After OnyX ran, the computer rebooted. I had to go to my chiropractor appointment, so I was walking out the store as it rebooted, and then while I was still in the store, the damn Airport turned off. I showed it to the Genius, and he said “Oh, that’s what you meant – oh, that’s definitely software”. I don’t understand why he didn’t understand what I meant at that point – when I described it the first time, the other guy behind the Genius Bar said “oh, you mean the pie wedge goes completely white”, and I confirmed that was exactly what I meant. I also don’t understand his conclusion. We’ll see.