Category: Geekery

Migration accomplished!

I’ve shutdown all three domUs on the old server and brought them up on the new server. So far, I’ve fixed a small issue with MySQL not starting up, which kept some of the web sites from starting up. But email appears to be flowing, the blog is up, news is up. Now to go through all the other services that should be running and make sure they are.

Upgrades, Updates

I upgraded WordPress to 3.0, and to celebrate I switched to the new theme. I’m going to to want to replace that banner picture at some point, but otherwise I think it went ok.

I “racked” my new colo box at LogicalSolutions.net/RackWire.com. Of course the first thing I did when I got home was discover that I had 15 un-installed security upgrades, including the kernels. Oh oh. I went ahead and upgraded, in spite of the nagging voice telling me that this didn’t go well last time. And the first thing I discovered that when it upgraded the kernels, it didn’t keep all the tweaked settings I had in /boot/grub/menu.lst, not even in menu.lst~. I also discovered that when I backed up the system before I brought it over to the facility, I only backed up the root partition, not /boot. Dammit. I’m currently rsyncing some files from the old box to the new one, but when that’s done, I’m going to reboot the new one, and cross my fingers and hope it will work. If it doesn’t, I guess I’ll see how well LogicalSolutions takes to me visiting my box for a few hours tomorrow.

Update: I rebooted my box with the new kernel and it actually booted! Hurrah!

More colo box setup

On the new box:

  • Boot it up
  • Shut down all three domUs with “xm shutdown”
  • Mount all three domU disks in /mnt/xen[1-3]

On the old box:

  • Set up an authorized key so that root on the new box can ssh into root on the old box
  • Make a snapshot of one of the disks with lvcreate -L 100G -s -n xen1-snapshot /dev/xen-space/xen1-disk
  • Mount it with mount /dev/xen-space/xen1-snapshot /mnt/xen1

On the new box:

  • Copy the changed files using rsync -aSuvrx --numeric-ids --delete -e ssh --exclude /lib/modules root@xen.xcski.com:/mnt/xen1/ /mnt/xen1

On the old box:

  • Unmount the snapshot
  • Remove the snapshot with lvremove /dev/xen-space/xen1-snapshot

What have I been up to recently?

Since the Tupper Lake race, I’ve only paddled the ski. I need to get used to paddling in waves, with the Rochester Open Water Challenge less than two weeks away. Tonight, for instance, Paul D and I did some surfing, but we also spent some time paddling up and down the shore – our theory was that we would experience waves from the side, which is the hardest to handle, but we were in shallow water so if we dumped (and I dumped a few times) it was no trouble to get back in the boat. I haven’t been paddling with my GPS much, so I don’t know what has happened to my training volume other than the feeling that it’s way down. Paddling out in the surf requires different muscles and it’s not particularly fast, so an hour or an hour and a half is about all I can stand, and I probably make less than 3 or 4 miles in that time.

I’ve also settled on a name for the ski. In the past, I named my Skerray “Mary Ellen Carter”, after the song by Stan Rogers, because it enabled me to “rise again”, and the Looksha was “Gideon Brown” after the song by Great Big Sea, because she can “punch ahead in any gale”. I called the Thunderbold “Anne-Marie” after the boat in Stan Rogers’ song “Acadian Saturday Night” because it has “wings on the water”. And so now I’m naming the ski Old Polina because I “fly a long like a song” in it. Or at least I hope to.

I had a great visit with my dad, step mother and kids this weekend. It’s great to see them, especially my daughters. They’re both maturing so well. I still worry about them, but I suppose that’s my job.

In other news, I’m still trying to finish setting up the replacement hardware. I’m experimenting with using LVM snapshots to be able to backup the domU partitions while they’re active – I think what I’ll do is make snapshots, rsync them over to the new server’s partitions, then delete them, and then shut down the domUs and rsync them again while they’re shut down, and then start up the new guys. By rsyncing once with snapshots, that should make the amount of time between shutting down the partitions and bringing up the new ones much faster. I’m also going to look into replacing my current rsync backup scripts to ones that use snapshots as well, because that way I never have to worry about inconsistencies in the file system, especially in the database engines.