3.7 pick list #16361
Description
this is a list of items in the release-1.7 branch we need to pick or rebase to pick up. Given the number of items, a rebase onto 1.7.x seems the most reasonable approach.
TBD (mostly in 1.7.8, some we already had picked):
-
#52933: Preserve leading and trailing slashes on proxy subpaths -
#52176: Use separate client for node status loop -
#52545: use specified discovery information if possible -
#43152: etcd3 store: retry w/live object on conflict -
#48394: GuaranteedUpdate must write if stored data is not canonical -
#53239: Correct APIGroup for RoleBindingBuilder Subjects
1.7.6:
-
#52385: Add --request-timeout to allow the global request timeout of 60 seconds to be configured. -
#52329: Use credentials from providers for docker sandbox image -
#52228: Fix discovery restmapper finding resources in non-preferred versions -
#52103: Fix proxied request-uri to be valid HTTP requests -
#52092: Fix resource quota controller panic -
#51969: Fixes an issue with APIService auto-registration affecting rolling HA apiserver restarts that add or remove API groups being served. -
#51954: Fix providerID update validation -
#51871: Use local JSON log buffer in parseDockerJSONLog. -
#51763: change SizeLimit to a pointer
1.7.5:
-
#51177: On AttachDetachController node status update, do not retry when node doesn't exist but keep the node entry in cache -
#51469: Enable finalizers independent of GC enablement -
#50862: Fix premature return in strategicmergepatch -
#50596: Allow unmounting bind-mounted directories. -
#51149: StatefulSet controller no longer attempts to mutate v1.PodSpec.Hostname or v1.PodSpec.Subdomain -
#51127: Fix initial exec terminal dimensions -
#49973: ScaleIO Volume Plugin - volume attribute updates -
#49767: make it possible to allow discovery errors for controllers
1.7.4:
-
#50623: csr: add resync to csr approver -
#50260: AWS: We never want to modify the globally defined SG -
#50513: AttachDisk should not call detach inside of Cinder volume provider -
#50250: fix data race in storage (during addition) -
#50240: Fix incorrect call to 'bind' in scheduler, Retry scheduling pods after errors more consistently in scheduler -
#50030: Make NodeRestriction admission allow evictions for bounded pods -
#50015: Correctly handle empty watch event cache
1.7.3:
-
#49768: fix pdb validation bug on spec(PR) -
#49781: fix check for nil value in interface for proxier health(PR) -
#49582: Fix issues for local storage allocatable feature(PR) -
#49578: Fix issue when setting fileysystem capacity in container(PR) -
#49592: Fix Operation names for subresources(PR) -
#49403: Use specified ServerName in aggregator TLS validation(PR) -
#49447: Do not spin forever if kubectl drain races with other removal(PR) -
#49455: Fix controller crash because of nil volume spec(PR) -
#49026: rebase gophercloud to support HTTP status 300 in pagination, so listingng Cinder v1/v2 API versions won't break(PR) -
#49254: Restore cAdvisor prometheus metrics to the main port(PR) -
#49381: glusterfs: retry without auto_unmount only when it's not supported(PR)
1.7.2:
-
#49155: make externalAdmissionHookConfigurationManager distinguish API disabled error(PR) -
#49067: Use local PX endpoint for mount, unmount, detach and attach calls(PR) -
#49089: Fix subPath existence check to not follow symlink(PR) -
#49069: Never prevent deletion of resources as part of namespace(PR) -
#48821: Correctly filter terminated pods in kubectl(PR) -
#48809: fix udp service blackhole problem when number of backends changes from 0 to non-0, local storage teardown fix(PR)
1.7.1:
-
#49047: hpa: Prevent scaling below MinReplicas if desiredReplicas is zero(PR) -
#48855: GCE: Fix panic when service loadbalancer has static IP address(PR) -
#48542: Ensure namespace exists as part of RBAC reconciliation(PR) -
#48490: Support NoSchedule taints correctly in DaemonSet controller(PR) -
#48261: Fix removing finalizer for garbage collector(PR)