AVAILABLE_WHEN_DONOR should be 1 as Fuel configures a non-blocking xtrabackup replication.
https://github.com/stackforge/fuel-library/blob/master/deployment/puppet/openstack/templates/galera_clustercheck.erb#L22
That would improve the UX with the full cluster reassemble replication time. Currently, when all galera nodes went down and cluster is being reassembled completely, the donor node will not accept any SQL connections - as HAproxy will not allow it - until the first node got synced from it and becomes ready for services. But there is nothing prevents the donor node to accept connections from the very beginning of the outgoing replication. So, the clustercheck script should report the donor as available ASAP - and that is what available_when_donor option does.
Fix proposed to branch: master
Review: https:/