... an imcompetant group of IT staffers (which BTW ne "my web staffers").

Our production webserver needed to be physically moved from point A to point B (two miles away). Our FreeBSD 5.3 was not officially supported at the time (only AIX, SUSE, and other flavors of the month were supported).

After the move was completed, my web staff noticed a developing trend that pertained to unexplicable system reboots. Many bad things were being said about our webserver's reliability, and we were told that we were using an unstable (and unsupported) FreeBSD build. "You must move to something reliable", we were told.

As directed, we investigated the available "supported" operating systems, and found a reasonable target for our web applications. In the meantime, our server had finally suffered a painful death.

My team insisted on doing a server post-mortem, only to discover that the chassis dual-power-supplies had been subjected to an overvoltage condition of 220VAC, rather than the nominal 110VAC.

Somehow (and miraculously via deus ex machina), a new power supply became available and was installed into our server.

To this day, (after over 200 days of uptime), our server has never been rebooted.

Hmmm... Fancy that!

$ uptime
  3:42pm  up 2040 day(s), 22:32,  2 users,  load average: 0.10, 0.07, 
+0.08
[download]

The poll missed the reason I last shut off my computer: facility mods. I had to change the light switch in my study. We only use 110V in the US, but I rather dislike the feeling of line voltage coursing through my body, so I pulled the breaker, and my computer is on the same circuit as the light switch.

It was due mostly to SATA not (as far as I'm aware) being fully hot-swappable (at least, not on BSD). Or, looking at things another way, it was due to the BIOS on another computer I was working on not being able to properly handle the SATA hard drive and the ATAPI CD drive at the same time.

See, we had this new computer, on which we wanted to install a certain OS distribution (a fairly special-purpose one, which happened to be based on (a rather old version of) slackware), but its BIOS had only two options for how to handle SATA: either it could be set up in "legacy" mode, in which case the SATA drive actually masked the IDE channel (and any drives on it) entirely, or else it could be set up in "native" mode, in which case the BIOS did not see the SATA hard drive at all. (A fully installed and booted OS with the proper driver could perhaps work around things and use the SATA drive in this mode, but the OS booted from the installation CD was not able to do that.)

So I brought both the hard drive and the CD to my FreeBSD workstation, which I turned off, plugged in the SATA drive there, booted back up, created a small partition on the SATA drive, copied over the data from the CD to there, then shut back down, took the SATA drive back out, restarted my FreeBSD workstation so it could resume its normal duties, took the new SATA drive back to the new system, and proceded with the installation. I had to make some modifications to the installation script to get it to use the hard drive partition in lieu of a CD drive, but it did work.

Actually, I had to do this twice, because for some reason the first time I did't quite get all of the package files from the CD copied onto the SATA drive, and did not become aware of this fact until later when I was partway through the install process.

So my FreeBSD workstation had downtime *twice*, due to a flaw in the BIOS of another computer. Why on earth the BIOS makers decided not to allow the SATA drive to be treated as an _additional_ IDE channel is entirely beyond me.

I suspect "downtime" is being used as the opposite of "uptime" -- as in, the last time your computer was off or rebooted.

I'm still looking for the "had to move" option. that's the reason i shut my work machine down today, 556 ... it's also the reason why i shut it down 556 days ago (i think it had over 2 years of uptime before that)

A couple moves ago, it occured to me that my server was plugged into a UPS that could power it for longer than the time it took to drive from my old apartment to the new one and briefly considered transporting them as a unit so that the box could stay up the whole time. But then I decided that was overly silly and just shut it down like a reasonable person since I didn't know how well the hard drives would respond to bumps in the road.

Ideally the reboot will only involve a minute or two of downtime, but it's still downtime. I mean, uptime gets reset to 0, after all...

(And how do you figure that 8 causes are OS-related? I only see two that I'd classify as such. (New kernel and BSOD.))

It was two things, %&#@&! blackout (which was also weather related), and the dead battery.

... nobody was using it anymore

That was actually true. Not *my* computer, but *a* computer in the computer room. We were making an inventory of what process was using what computer, and if it needed moving to the brand new machine that had been installed, and in this process, nobody was able to tell why one of the machines was still powered on. Last login showed 4 month ago.

Needless to say noone actually misses this rotten AIX 4.3 box

You didn't start it smoke testing perl?

To be honoust, that was the only thing it still was doing, see here

But, as I hate AIX, and have no tuits or intention to fix any failure that would come out of its smoke results, and noone else still (on p5p) running this archaic OS, I saw no good reason to leave this machine consuming power just to smoke blead.

I still smoke HP-UX 10.20, only powered it down for my holiday, because we need the phisical space for some infrastructural reorganizations, but other than for that AIX box, I still do something useful with this box


Enjoy, Have FUN! H.Merijn

I've actually kicked network connections loose (why are they located exactly where I put my feet when sitting at my workstation in my cramped, uncomfortable cubicle?).

At one work place, we actually had one of the sysadmins kick the plug out of the wall. It did little to improve my opinion of sysadmins.

And in one of my jobs, we had a customer show off his brand new UPS by pulling the plug. He wanted to pull the UPS's plug out of the wall socket; instead he pulled the computer's plug out of the socket on the UPS. Ooopsie!

I can't help you with network connections so much (it just makes me miss the office when our network jacks and power were above the desks -- the room had been a former machine room, and they kept the power high, and they didn't want to run the network across the AC)

But, I've kicked my share of power strips, and seen other people do the same, so I'll mention a trick that one of my former co-workers showed me:

Kick-proofing your power strip's on/off switch

1. Obtain the lid from a Gatorade or similar sized lid (about 1.5"/4cm inner diameter).
2. Using a box cutter or other utility knife, cut a rectangular hole in the top of the lid, large enough to get your finger through.
3. Orient the rectangular hole over the rocker switch on the power strip, and secure with duct tape.
4. Cut the duct tape away to expose the hole.

If you see no reason to ever turn off your power strip, then you only need steps 1 and 3, and no need to orient the lid.

It took me a week to recover most of the partitions back, but for some reason, I couldn't rescue data from two partitions, not in easy way (I'm not giving up yet though). Now, none of my two Linux installations are fully working in the same state before the disaster.

I'm using another laptop now, reinstalling a distro, and in the process of setting up everything to build my development environment: web server, rdbms, CPAN, etc, you know.

A very very very expensive lesson for me: before doing high-potential-destructive-operation, think twice, no, no, think three, four, or five times. Well, the more you think the better. Many rescue tools are available out there, and they're getting better and smarter. But things are not always going smooth.

Two random hints:
1. Access your disk with parted and use the "rescue" command to find the partition locations.
2. Backup your partition table by copying it with dd if=/dev/hda ... to a safe place.

My last downtime was due to dual-booting (between Mepis Linux and Gentoo Linux... yeah, I'm strange). Prior to that was weather-related power outage. I'm too cheap to buy a UPS.

Required for OS security updates, that our management required.

... A new UPS.

The very device designed to prevent down time failed, turns out that the old UPS held as much charge as a length of power cable can -_-

Actually, my Linux box is getting cantankerous in it's old age and in a typical case of "the shoemaker's children" I haven't gotten around to upgrading the OS on it. For that reason I have to reboot the damn thing every day which is done out of cron at noon.

On my shopping list for this old beast is:

1. A new hard drive (old one is showing bad sector errors in /var/log/messages)
2. More RAM. I have so much stuff running on there the poor beast is IO bound swapping stuff in and out as well as paging according to vmstat
3. A new fan. The chassis fan has given up the ghost.
4. A 17" LCD monitor since I'm still using that old CRT that I fished out of a trash can 5 years ago!

Someday the poor beasty is going to get replaced by a new machine that I build from scratch and demoted to being my workbench machine and relegated to being a diskless workstation. I just don't have the $$ to spend right now.

Apart from rebooting so that security upgrades take effect (which I don't really count as downtime), my last downtime was when I messed up my last OS upgrade. It took me 4 days to recover.

Power supply pooped out.

I was sitting in my dorm at UNC and i heard a loud POP and saw a bright flash. Tried to turn it back on and nothing happened. DAMN!

• the maid.

Ours, who comes once a week, was looking for somewhere to plug the iron in....

This was, um... about 71 days, 4:58 ago ;)

(I've also had to ban her from "dusting" my cable modem - as she kept resetting it in the process :/)

Reminds me of an episode of a Brittish detective, I think it was Frost, where people died in a hospital because the cleaning lady pulled out plugs of vital medical equipment to plug in the vacuum cleaner. It was never noticed before, because she always put the plug back in when she was done cleaning.


Enjoy, Have FUN! H.Merijn

powerful electromagnets, or heat (probably heat), or shoddy craftsmanship

*shakes head* Sorry, I spent last evening talking to my in-laws.