CON-648 |
csi.k8s |
Implement basic topology in the CSI driver |
StorageClasses may now be decorated with topology requirements and use volumeBindingMode: WaitForFirstConsumer |
CON-2508 |
csi.k8s |
Post reboot and drained worker nodes are not cleaned up properly unless manually tidied up |
Safely removing stale mount paths or devices attached to a compute node. |
CON-3071 |
csi.k8s |
CSI node monitor support |
A node monitor is now patrolling the block devices to ensure they're healthy. |
CON-2137 |
csi.k8s |
Run filesystem check before mount |
Checking for filesystem corruption and optionally repair it with the fsRepair StorageClass parameter. |
CON-3082 |
csp.3par |
Cloning RWO to RWX does not apply multiInitiator in CRD |
While cloning any RWO volume, adding the multiInitiator flag if the target volume access mode is RWX. |
CON-3088 |
csp.3par |
MultiInitiator does not get set during volume import |
Adding the "multi_initiator" flag if the target volume access mode is Block RWX during the volume import. |
CON-3090 |
csp.3par |
Change importVol parameter to importVolumeName for 3PAR CSP |
Making the parameter consistent with the CSI/CSP naming convention. |
CON-3051 |
csi.k8s |
CHAP need to use a Secret instead of a ConfigMap |
Moved the CHAP user credentials to CHAP secret instead of populating a Helm value parameter in a ConfigMap. |
CON-3099 |
csi.k8s |
PVs created in CSI Driver version 2.4.2 or lower version lose CHAP details in 2.5.0 |
Support backward compatibility for existing PVs using CHAP authentication already. |
CON-3080 |
csi.k8s |
CHAP enables iSCSI initiators trying to establish session and fail indefinitely leading to flood of events followed by FW cores and iostack panics. |
Not deleting the host definition if the volume is using CHAP authentication with a Helm chart parameter disableHostDeletion. |
CON-2867 |
csp.nimble |
Unable to import offline vVols into Kubernetes |
Set agent_type to "none" during forceImport. |
CON-3064 |
csi.k8s |
Implement nfsResourceRequestCpuM and nfsResourceRequestsMemoryMi |
Added request limits for CPU and Memory resources. |
CON-3106 |
csi.k8s |
The hpe-csi-node Pods are consuming more than 100GiB of memory on nodes |
Implemented resource requests and limits on all containers in the HPE CSI Driver. |
CON-3094 |
csi.k8s |
CSI Operator controller manager gets OOMkilled |
Doubled the default memory limit. |
CON-3087 |
csi.k8s |
Use Red Hat branded Operator SDK for OpenShift |
Replaced references in HPE CSI Operator with images from registry.redhat.io |
CON-3092 |
csi.k8s |
ClusterRoles conflicts with other CSI drivers |
Prefixed ClusterRoles with 'hpe-csi'. |
CON-2930 |
csi.k8s |
All images needs to be part of the Operator manifests to allow offline install via digest |
Added all images to the Helm chart used by the CSI driver. |
CON-3085 |
csi.k8s |
Refactor the Helm template logic to parameterize all images |
Added "images" to the Helm chart. |
CON-3086 |
csi.k8s |
NFS Server Provisioner is hardcoded in the CSI driver |
Moved the default NFS server image to a runtime changeable parameter. |
CON-3003 |
csi.k8s |
Ubuntu gets a deprecated multipath.conf |
Changed getuid method to ID_SERIAL. |
CON-3084 |
csi.k8s |
Allow arbitrary NFS node label values |
Implemented an nfsNodeSelector StorageClass parameter. |