Enabling Non-zero Value for "live_migration_progress_timeout" in Nova Results Live Migration Failure

Problem

• Live migration fails with the below errors after waiting for some time for completion.

WARNING nova.virt.libvirt.migration [[req-id] [user@domain.com] Production] [instance: [instance-uuid]] Live migration not completed after 896 secERROR nova.virt.libvirt.driver [[req-id] [user@domain.com]  Production] [instance: [instance-uuid]] Live Migration failure: operation aborted: migration out job: canceled by clientERROR nova.virt.libvirt.driver [[req-id] [user@domain.com] Production] [instance: [instance-uuid]] Migration operation has aborted

• Setting a non-zero value of live_migration_progress_timeout in Nova results in live migration failure.

Environment

• Platform9 Managed OpenStack - All Versions
• Nova

Cause

The compute service aborts a live migration when it detects that memory copy is not making progress based on the value set for live_migration_progress_timeout.

Resolution

1. Set the value to zero for live_migration_progress_timeout in /opt/pf9/etc/nova/conf.d/nova.conf or /opt/pf9/etc/nova/conf.d/nova_override.conf on the host.

xxxxxxxxxx

live_migration_progress_timeout = 0

2. Setting this parameter to value zero disables progress timeouts.

3. This value is set to 0 by default in Platform9 Managed OpenStack v4.0 and higher.

Additional Information

Refer to the OpenStack documentation for more details on config options in Nova.