Version 5 (modified by, 10 years ago) (diff)


Emergency Stop Procedure

Emergency stop procedures are intended for allowing an operator to shut down part of or an entire slice in case the slice poses danger to one of the hosting campuses, the testbed, or the public Internet.

For the emergency stop to succeed some of the following information must be provided:

*Name of the slice
*Owner of the slice

The following is a summary of available options for shutting down all or elements of a particular slice based on these inputs:

  • Shut down slice or slivers through the tool.
    • Required Input: slice name
  • Shut down VM slivers on a specific rack using OpenStack
    • Required Input: VM UUID from the manifest

Shutdown via Omni tool

  1. Anyone with the proper slice credentials, can run:
  1. delete 'slicename' -a aggregate-name (AM API V3) or
  2. deletesliver 'slicename' -a aggregate-name (AM API V2)

Shutdown via OpenStack Horizon Administration GUI

  1. From a web browser go to http://ControlNodeIPaddress/horizon
  2. Login with the admin password from /etc/novarc
  3. Click on the Admin Tab
  4. Click on Instances
  5. Click on Instance names to find matching UUID
  6. Go back to Instances and choose the correct instance and Choose Terminate Instances or you can click on the more button and choose terminate from there.

Shutdown via OpenStack Command Line

  1. Login to the control node
  2. 'su' to user 'gram'
  3. Source the novarc file /etc/novarc
  4. Run 'nova list' command
  5. Verify that a VM with a matching UUID is running
  6. Run 'nova delete <UUID>' to shut down the VM

If the shutdown is performed via OpenStack, it is desirable to then close the slice or the particular sliver reservation for the VM through, if possible.