Skip to content

v0.7.0
Compare
Choose a tag to compare
@davidz627 davidz627 released this 05 Mar 01:23
· 1429 commits to master since this release
v0.7.0
fb027cc
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

Image: gcr.io/gke-release/gcp-compute-persistent-disk-csi-driver:v0.7.0-gke.0

Changelog Since v0.6.0

This is a fairly large release with some serious fixes and new functionality. There are also some breaking changes that will require intervention and special steps if upgrading from a previous version of the driver.

This is also my (davidz627@) final release of the driver.

adieu.

Changes with Action Required

  • Adding PodSecurityPoliciy to allow csi-gce-pd-node in clusters with policies enabled.
    IF LOCAL PSP MANIFEST PATCH IS USED PLEASE BEWARE THAT YOU WILL NEED TO DELETE LOCAL CHANGES AND USE THE UPSTREAM (#448, @ffilippopoulos)
  • BREAKING CHANGE: All deployment objects in setup-cluster.yaml have been renamed. When deleting the deployment using ./delete-driver.sh, make sure to use specs from your previous deployment version to ensure the correct objects are cleaned up. (#405, @verult)

New Features

  • Add GET_VOLUME_STATS Node Service Capability and implementation for getting stats for volume (#406, @davidz627)
  • ValidateVolumeCapabilities validates that the given volume conforms to all capabilities in the request. Validation of existing volumes during inserts also improved to check all parameters. (#467, @davidz627)
  • It is now possible to disable the controller service by setting --run-controller-service=false. Similarly, it is possible to disable the node service by setting --run-node-service=false. The latter enables running the controller server of the GCE PD driver separately/outside of the cluster it is serving. Also, if both project-id and zone are specified in the GCE cloud config then the controller server does no longer try to contact the GCE metadata service. (#449, @rfranzke)
  • Add support for formatting and mounting an XFS filesystem (#447, @davidz627)
  • Add a blanket toleration to the Node Daemonset of the driver deployment so that it can be deployed on all nodes (#417, @davidz627)
  • Adds LIST_VOLUMES and LIST_VOLUMES_PUBLISHED_NODES capabilities with respective functionality (#392, @davidz627)

Bug Fixes

  • Fixed bug where ControllerExpandVolume was returning incorrect size when disk was already the requested size or larger (#462, @davidz627)
  • Set volume limits to 15 only for machine-types: "f1-micro", "g1-small", "e2-micro", "e2-small", "e2-medium". Limit is 127 for all others (#455, @davidz627)
  • Changed deployment of Controller and Node components to use hostNetwork for compatibility with GKE Workload Identity (#436, @davidz627)
  • During NodeStageVolume run udevadm --trigger to fix device symlinks if device path is not found or device path points to the wrong device (#459, @davidz627)
  • Bump external-snapshotter version to v1.2.2 for fix of CVE-2019-11255 (#434, @davidz627)

Other Notable Changes

  • Update driver base image distro to debian-amd64:v2.0.0 and build with go v1.13.4 (#439, @davidz627)
  • Mounting an unformatted volume with an fstype as read-only now throws a more descriptive error (#458, @davidz627)
  • Remove explicit stripping of secrets from RPC request/response logs since the driver doesn't accept secrets for operations (#428, @davidz627)
  • Improve driver logs to log success in all paths as well as logging additional useful information (#409, @davidz627)
Assets 2
Loading