I'd think that rather than the resource tracker dealing with failed migrations, we'd clean those up when the failed compute node is recovered, something like what we have here:
https://github.com/openstack/nova/blob/569e463f02b0a631d341b06849217657a22c4569/nova/compute/manager.py#L636
I'd think that rather than the resource tracker dealing with failed migrations, we'd clean those up when the failed compute node is recovered, something like what we have here:
https:/ /github. com/openstack/ nova/blob/ 569e463f02b0a63 1d341b068492176 57a22c4569/ nova/compute/ manager. py#L636