forked from rook/rook
Sébastien Han
7788901a88
Now Kubernetes will perform liveness checks on mon, mds and osd daemons. The command will: * call the socket (check for existence) * execute a command and check the return code (success if 0) This handles the case where the daemon is stuck locally and unresponsive. It's unlikely but not impossible. These checks bring more robustness to the implementation. rbd-mirror and nfs have been leftover for the following reason. The rbd-mirror socket name is different from other daemons (could be fixed though): /run/ceph/ceph-client.rbd-mirror.a.1.94362516231272.asok also, the command to call would need to be changed from "status" to "rbd mirror status" so we can keep this for a later. The nfs ganesha has no socket only a PID file which doesn't mean much. No PID means the process does not run so Kubernetes will already handle this and the pod will crash loop. Signed-off-by: Sébastien Han <seb@redhat.com>
2.5 KiB
2.5 KiB
Major Themes
Action Required
Notable Features
- Added K8s 1.17 to the test matrix and removed K8s 1.12 from the test matrix.
Ceph
- OSD refactor: drop support for Rook legacy OSD, directory OSD and Filestore OSD. For more details refer to the corresponding issue.
- OSD on PVC now supports a metadata device, refer to the cluster on PVC section or the corresponding issue.
- OSD on PVC now supports PVC expansion, if the size of the underlying block increases the Bluestore main block and the overall storage capacity will grow up.
- Ceph Nautilus 14.2.5 is the minimum supported version
- OSD on PVC doesn't use LVM anymore to configure OSD, but solely relies on the entire block device, done here.
- Specific devices for OSDs can now be specified using the full udev path (e.g. /dev/disk/by-id/ata-ST4000DM004-XXXX) instead of the device name.
- OSD on PVC CRUSH device storage class can now be changed by setting an annotation "crushDeviceClass" on the "data" volume template. See "cluster-on-pvc.yaml" for example.
- Rook will now refuse to create pools with replica size 1 unless
requireSafeReplicaSizeis set to false. - CSI drivers can now be configured using "rook-ceph-operator-config" ConfigMap. ConfigMap can be used in mix with the already existing Env Vars defined in operator deployment manifest. Precedence will be given to ConfigMap in case of conflicting configurations.
- Rook Ceph cleanupPolicy will clean up the dataDirHostPath only after user confirmation. For more info about CleanUpPolicy read the design as well as the documentation cleanupPolicy
- Rook monitor, mds and osd now have liveness probe checks on their respective sockets
EdgeFS
YugabyteDB
- Master and TServer pods for YugabyteDB will have resources requests and limits specified as per YugabyteDB recommendations. This will help avoid the soft/hard memory limit issue.
Cassandra
- Added JMX Prometheus exporter support.
Breaking Changes
Minio
- The minio operator was removed from Rook