Miguel was able to reproduce this by spawning a VM and then destroying it, top would then show the agent in a tight loop. I was able to do the same and found that even with my fix that got merged, the agent could still go into a tight loop. Thomas' change is better, https://review.openstack.org/#/c/549331/ - I just updated it based on additional testing and have proposed a revert of my change that merged.
Miguel was able to reproduce this by spawning a VM and then destroying it, top would then show the agent in a tight loop. I was able to do the same and found that even with my fix that got merged, the agent could still go into a tight loop. Thomas' change is better, https:/ /review. openstack. org/#/c/ 549331/ - I just updated it based on additional testing and have proposed a revert of my change that merged.