1.4.0 Aster
This is the first release in the Aster 1.4 series.
Notable changes
Include Asklepios auth-healing service for non-graceful node shutdown. It detects a node failure and quickly moves pods from that node to other nodes to minimize service failure.
Enhance OpenStack HA failover.
Rebuild openstack yoga images to update oslo.messaging (upgraded to 14.2.4).
Set replicas to 1 for cinder-scheduler, nova-scheduler, nova-conductor, and placement services for quick failover by Asklepios.
Enhance rabbitmq connection settings for openstack services
Burrito Changelog
remove cluster config in cinder conf; (jijisa@iorchard.net)
update vars.yml.sample; (jijisa@iorchard.net)
remove jinja2_native; (jijisa@iorchard.net)
update kubelet-csr-approver variables; remove coredns nodeselector; (jijisa@iorchard.co.kr)
add jinja2_native = true in ansible.cfg.sample; (jijisa@iorchard.co.kr)
fix(openstack): set show_multiple_locations for cinder backends in glance to support volume-to-image operation; (jijisa@iorchard.co.kr)
fix(openstack): change heat image to ubuntu_jammy image; (jijisa@iorchard.co.kr)
upgrade openstack images to ubuntu_jammy in scripts/images.txt; (jijisa@iorchard.net)
update placement images to ubuntu_jammy; (jijisa@iorchard.net)
update nova images to ubuntu_jammy that includes oslo.messaging upgrade to 14.2.4; add heartbeat_in_pthread: true; (jijisa@iorchard.net)
update neutron images to ubuntu_jammy that includes oslo.messaging upgrade to 14.2.4; (jijisa@iorchard.net)
update keystone images(ubuntu_jammy); add heartbeat_in_pthread: true; (jijisa@iorchard.net)
update cinder image; cinder-scheduler replica=1; add volume cluster config; add heartbeat_in_pthread: true; (jijisa@iorchard.net)
change rabbit_interval_max to the default value 30; (jijisa@iorchard.net)
adjust nova-conductor/scheduler liveness/readiness parameteres; (jijisa@iorchard.net)
increase cinder-api liveness/readiness timeoutSeconds to 3 seconds; (jijisa@iorchard.net)
change rabbitmq readiness script to use check_local_alarms; (jijisa@iorchard.net)
change rabbitmq liveness script to use check_running; (jijisa@iorchard.net)
upgrade asklepios to 0.2.0; (jijisa@iorchard.net)
update vars.yml.sample; (jijisa@iorchard.net)
change placement-api replicas to 1; (jijisa@iorchard.net)
change nova-{scheduler,conductor} replicas to 1; (jijisa@iorchard.net)
make barbican installation optional; (jijisa@iorchard.net)
adjust nova liveness/readiness time; (jijisa@iorchard.net)
feature(aster): modify transport_url in openstack config; (jijisa@iorchard.net)
fix: add transport.cleanup in health-probe.py; (jijisa@iorchard.net)
add asklepios variables in vars.yml.sample; (jijisa@iorchard.net)
add asklepios image url in images.txt; (jijisa@iorchard.net)
add asklepios image and patch in genesis registry; (jijisa@iorchard.net)
add asklepios role in patch playbook; (jijisa@iorchard.net)
fix(gnsh): update gnsh script to try to uncordon the node multiple times until kube-apiserver is ready; (jijisa@iorchard.net)
fix(gnsh): do not run gnsh role when there is only one control node; (jijisa@iorchard.net)
add calico images in genesis registry; (jijisa@iorchard.net)