Brad Fitzpatrick (bradfitz) wrote in lj_maintenance,
Brad Fitzpatrick

Good news

It seems we only post bad news about servers being dead here, so today I thought I'd post some good news, and a little heads-up.

We've been moving tons of users to our new databases, which are finally in production after months of vendor problems.

If you see a message that your journal is in read-only mode, that's probably what's happening... your journal is locked while it's being copied to the new servers. It shouldn't last more than a couple minutes. (not that the copies take that long, but we lock the accounts when they're "on deck" for an upcoming move, to let the account quiesce)

These new servers are our first 64-bit pair (so they can use way more memory), they have twice as many disks as previous servers (which makes them both bigger and faster), they're running InnoDB, and they're absolutely kicking ass. With about 75k active users on them, you can't even tell they're doing anything, they're so idle. That number will rise quickly over the next week.

If you're curious, keep an eye on the Where am I? page. More than likely you'll end up on the new DB cluster ("Roastbeef") or either Porkchop/9 or Madcow/9 at some point.

The plan is to empty all the old databases, upgrade and reconfigure them, then put them back in service.

Recent Posts from This Community

Comments for this post were disabled by the author