OK, time to gracefully reboot the box again.

I've shed a few users off the machine, and increased the timeout for devices connecting to Xen's backend disk device driver. That should help eliminate the majority of the "failed to get domid" error messages. However:

http://lists.xensource.com/archives/htm ... 00170.html

Once the machine gets into this state, no new block devices can be attached. That's clearly (to me, anyway) a bug in Xen's backend block driver, but what I suspect is that because of the small timeout value (10 seconds), something goes awry to cause that bad state.

I'm going to shutdown the nodes and reboot the host in a few minutes.

-Chris

There once was a cake from Stafford.
He found Xen to ease the discord.
He lost his domid, so he tried again
But the daemon acted of its own accord.

Increasing the timeout for device attaches seems to have worked. There weren't any "failed to get domid" errors in any of the boots this time.

However, there's still the performance aspect I'm fighting with. Still digging...

-Chris

My node was "Powered Off". The job queue says there was a successful boot, but I don't see any other signs of it.

It's coming up now... slooowly...

I lost my domid!! help! I've looked everywhere for it. I must have left it at the library.

Or maybe my dog ate it.. he has been acting weird today. Yea, thats it.. my dog ate my domid.