Things always come in threes: buses, computer problems, and lists of things that come in threes.
At the beginning of the week, oh so long ago now, I decided to modify my email config on the grounds that if it ain’t broke I should tinker with it until it is. I already use my own Linux box as an SMTP server and IMAP server, but the SMTP server isn’t public. I wanted to open it up to authenticated users only (me), so I could use it for sending email while not at home.
Long story short, I broke the cardinal rule and changed too much at once. Net result: incoming email started to disappear into the bit bucket, no idea why (since it had worked for a time). I reverted the changes and on inspecting various log files thought that my libraries might be a bit wonky. Hell, it couldn’t hurt to run revdep-rebuild just to see.
I naturally ran it first with the don’t-actually-do-it option. And it proceeded to not do it quite magnificently – it came to a shuddering halt part-way through the discovery process, while I was writing an email (now that I could do that again). I only noticed the problem when Thunderbird began to whine about its inability to save a draft on the server. It was unable to do so, it transpired, as the server was wedged solid. The console was unresponsive. Caps Lock didn’t.
Oh, I thought, never mind; six months uptime ain’t bad. I powered down and waved the Dyson at it for a few minutes, since the various fans were doubling up as dust factories. Then I powered up.
Well, I pressed the button. Current neglected to flow, and continued to stubbornly resist no matter how many times or how hard I pressed the power button.
Oh, I thought again. I may even have said it out loud.
I checked the things everyone forgets to check: plugged in, switched on, socket working, etc. I wiggled connectors. I banged things. Following a flowchart I found on the innertubes I unplugged disks and cards. Nothing. Knackered power supply or motherboard, the flowchart said – try a replacement power supply first.
Luckily I still have an old server liberated from Tarantella, willing to donate organs for medical purposes. (It’s an old Gateway PC that I think came from the Watford office.) I extracted the PSU and plugged it like an artificial external heart into my sick server, leaving the old PSU where it was (I imagine you’re not supposed to do this, but hey).
It worked! The server came alive, booting as far as it could before it noticed it didn’t have any disks to play with. I powered down again and rejoiced.
Next day I unplugged both temporary and original PSUs. The temporary one had a clunking great fan in just the wrong position for my motherboard – no good as a permanent replacement. I nipped to PC World with the original PSU and found one that would fit, getting a Man to double-check that it would be suitable (what do I know about PSUs?).
Brought it home, plugged it in, turned it on. Brilliant, back to normal.
Half an hour later it wedged again, in the middle of a build. Thankfully cycling the power worked this time but now the CPU fan was in trouble. Hardware guru that I now am, I surmised that if you can see the fan rotating then it may as well not be. Both wedges, it seemed, were caused by a roasty-toasty CPU – the first one pushing the PSU over the edge in a glorious cascade of failure. (This is what you get when you run a six-year-old Evesham desktop PC 24×7 for a year.)
Back to PC World for a new fan which, amusingly to me, screwed right into the heatsink. Yes, that’s better, and quieter too.
I now sit with fingers crossed, waiting for the third thing. I want to do a back-up, but that’ll be it. Maybe I should rebreak my email first.