Created attachment 686623 [details] Failure logs on controller and compute node Description of problem: nova scheduler doesn't report failure of selected host and reason it failed I have change nova.conf vlan_interface to non exist one Lanuch new VMs will fail on the compute node . no event is seen in /var/log/nova/scheduler.log on controller, when enabling debug some info is seen but doesn't explain what caused it this is problematic , in case some compute nodes are used and one fails to start VM from some reason , Error/Warrning event must be sent by nova/scheduler.log Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. create a setup with controller and two compute nodes 2. check VMs are deployed on both of them ( nova list , and nova show <VM ID > to see both hosts are used for VMs 3. change interface used by private network to interface that doesn't exist 4. reboot the host 5. launch new VMs 6. check logs in both compute node and nova/scheduler.log on controller Actual results: Attached logs from this example Expected results: Additional info:
The scheduler doesn't know why it failed. In this case, it's a problem local to the compute node, and you do get an error that gives a hint as to what the problem is related to in the compute log.