yahoo-eng-team team mailing list archive

[OpenStack Infra <1518848@xxxxxxxxxxxxxxxxxx>]

Reviewed:  https://review.openstack.org/248539
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=f3a2f8fc1048a70426b44e614790c2d8db431bbc
Submitter: Jenkins
Branch:    master

commit f3a2f8fc1048a70426b44e614790c2d8db431bbc
Author: ChangBo Guo(gcb) <eric.guo@xxxxxxxxxxxx>
Date:   Mon Nov 23 14:16:23 2015 +0800

    Fix reclaim_instance_interval < 0 never delete instance completely
    
    CONF.reclaim_instance_interval is interval in seconds for reclaiming
    deleted instances. It should take effect only when value is greater
    than 0. But current when set it with -1 for example, it will pass check
    if we should soft delete an instance, but can't pass check whether should
    clean soft deleted instance, then never have chance to clean up a soft deleted
    instance and release the quota.
    This commit sets CONF.reclaim_instance_interval's minimum value to 0.
    
    Closes-Bug: #1518848
    
    Change-Id: Idd8faad1a40f8898a9471504e664c3676b1b0e06


** Changed in: nova
       Status: In Progress => Fix Released

-- 
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/1518848

Title:
  set reclaim_instance_interval < 0, never delete instance completely.

Status in OpenStack Compute (nova):
  Fix Released

Bug description:
  CONF.reclaim_instance_interval  is interval in seconds for reclaiming deleted instances.  It should takes effect only when value is greater than 0.    current if we set it with -1 for example .  It will pass the check of ' if CONF.reclaim_instance_interval:' and do a soft_delete.   See https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/servers.py#L752
      def _delete(self, context, req, instance_uuid):
          authorize(context, action='delete')
          instance = self._get_server(context, req, instance_uuid)
          if CONF.reclaim_instance_interval:
              try:
                  self.compute_api.soft_delete(context, instance)
              except exception.InstanceInvalidState:
                  # Note(yufang521247): instance which has never been active
                  # is not allowed to be soft_deleted. Thus we have to call
                  # delete() to clean up the instance.
                  self.compute_api.delete(context, instance)
          else:
              self.compute_api.delete(context, instance)

  We do reclaim instance in  https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L6192
      def _reclaim_queued_deletes(self, context):
          """Reclaim instances that are queued for deletion."""
          interval = CONF.reclaim_instance_interval
          if interval <= 0:
              LOG.debug("CONF.reclaim_instance_interval <= 0, skipping...")
              return

  That means we never have chance to clean up a soft deleted instance
  and release the quota.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1518848/+subscriptions