Chapter 5. Notable bug fixes
Red Hat OpenShift Container Storage 4.4 introduces the following notable technical changes:
Table 5.1. List of fixed bugs
Previously, when a worker node was down, the operator was blocked from responding to CR updates such as upgrades, adding storage, or creating new pools. This issue has been resolved, and the operator works as expected.
When 'portable: false' was used in a StorageCluster CR; for example, Local Storage infrastructure, it was ignored and the PVC ID was used as the name for the host CRUSH bucket in Ceph. With this fix, "portable: false" is honored in the CephCluster and CRUSH bucket uses hostnames instead of PVC IDs for name of the host.
When an underlying disk of an OSD failed, the cluster always remained in WARNING state after data re-balancing as there was no way to remove the OSD from the cluster. The cleaning up of the failed OSD has now been simplified with a job that the administrator can launch.
Previously, the reported size by Red Hat OpenShift Container Platform for a node was slightly less than the actual size (64 GiB free memory) resulting in validation failure. For example, AWS M5.4xlarge machine had 16 core and 64 Gib memory of RAM but memory size reported by OpenShift node API was 61.xx GiB. Hence, an unexpected warning message was displayed for the expected configuration.