LJ

[cdybedahl]

So LJ is back again and we can all return to our usual form of timewasting. Nice.

Actually, I'm slightly impressed that they got the whole thing up again in only about a day. More than one hundred machines, most of them clustered database servers, all going down hard to power loss at the same time and under production load... That's the sort of thing that makes sysadmins wince.

Comments

[mistraltoes.livejournal.com]

Yeah, as much as I wish it had been faster, I was impressed, too.

[hawkeye7.livejournal.com]

I was impressed too. When you lose power, I always expect to lose disks, even if they are carefully powered down rather than, as in this case, going down hard when the lights go out. I would not be surprised to lose ten disks out of every hundred on a power failure. For some reason old disks keep spinning for ages but fail to start up again. Disks these days are more reliable than the old but, being larger, are harder (and take longer) to recover. Twelve hours is not unusual for a database server.