Hi Barton, Thanks for your clarification for current VIM design! Hi Peng, Could you retest and double check it again? Need to wait for enough time as mentioned in Barton’s comment.

From current log, there are 2 VMs already resumed on ctrl-1 after it reboot.

The VIM is behaving correctly in this case. When a host with instances on it goes offline, the behaviour is as follows: - As long as the host is disabled, attempt to evacuate the instances from the host (one at a time). - When the host goes enabled (e.g. after it reboots) do not do any further evacuations (this could lead to an instance running on multiple hosts). - Once the hypervisor on the host goes enabled, recover any failed instances by rebooting (or rebuilding if the reboot fails). Thanks! Zhipeng From: Wensley, Barton <Barton.Wensley@windriver.com> Sent: 2020年1月11日 1:29 To: Liu, ZhipengS <zhipengs.liu@intel.com> Cc: starlingx-discuss@lists.starlingx.io Subject: RE: [Starlingx-discuss] Discuss about one VM evacuation failure issue Zhipeng, I took a look through the logs and the VIM is behaving correctly in this case - I added my analysis to the LP. Bart From: Liu, ZhipengS [mailto:zhipengs.liu@intel.com] Sent: January 9, 2020 9:23 PM To: Wensley, Barton Cc: starlingx-discuss@lists.starlingx.io<mailto:starlingx-discuss@lists.starlingx.io> Subject: [Starlingx-discuss] Discuss about one VM evacuation failure issue Hi Bart, I have been working on one OpenStack related LP below. https://bugs.launchpad.net/starlingx/+bug/1857462 You can get detail information in LP comments. Root cause was found after analyzed related log, it seems need a fix in NFV side. As you are NFV expert, I’d like to get your comment before I can work out the solution. Thanks! zhipeng