yahoo-eng-team team mailing list archive

[Sean Dague <sean@xxxxxxxxx>]

** Changed in: nova
       Status: Incomplete => Invalid

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1208364

Title:
  The compute service seems not been notified/casted after long run

Status in OpenStack Compute (Nova):
  Invalid

Bug description:
  The tester was running a workload of around 1000 VMs for about 96 hours. The first 2 days, everything was fine, all VMs deployed were ACTIVE. Starting on the 3rd day, some of the VMs deployed (about 3%) stuck in BUILD state. They never change to ACTIVE or ERROR. He checked these VMs, most of them (all but 2) were scheduled to go to one of the hosts cn24. When he checked cn24, its quantum and compute services were both up and running.
  On the host side, the compute logs don't have any entries regarding this VM, it looks like the host never got the notification of spawning this VM. Somehow, there seems to be a disconnection between the controller and host, although the host is showing up and running from the controller.
  After he restarted the network and compute services on the host, subsequent VMs deployed to this host are ACTIVE. So it seems maybe the compute service on the host has some problem? although it shows it is running, but it actually isn't fully running?
  For the experts here, please share your opinions, thanks in advance!

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1208364/+subscriptions