Addressed issues¶
Issues addressed in the MKE 3.6.11 release include:
[MKE-10835] Fixed an issue wherein Gatekeeper Pods were frequently entering
CrashLoopBackOff
due to the absence of an expected CRD definition.[MKE-10640] Fixed an issue wherein a vulnerability in etcd library
go.opentelemetry.io/contrib/instrumentation/google.golang.org/grpc/otelgrpc
allowed a memory exhaustion attack on the grpc server. The patch is applied on top of etcd version 3.5.10 and includes the fix for CVE-2023-47108. No other changes from etcd 3.5.10 are included.[FIELD-6835] Fixed an issue wherein
ucp-cluster-agent
continually restarted during upgrade to MKE 3.6.10.[FIELD-6695] Fixed multiple issues caused by large etcd instances:
The addition of a second manager sometimes caused etcd cluster failure when the etcd storage size was of substantial size.
Cold starting of MKE clusters would fail whenever the etcd storage size exceeded 2GB.
Simultaneous restarting of manager nodes would fail whenever the etcd storage size exceeded 2GB.
The etcd storage usage indicator in the MKE web UI banner was not accurate.
[FIELD-6670] Fixed an issue wherein CNI plugin log level was not consistent with the MKE log level setting.
[FIELD-6573] Fixed an issue wherein kubelet failed to rotate Pod containter logs.