Skip to content

1.12.0-rc2

Pre-release
Pre-release
Compare
Choose a tag to compare
@aanm aanm released this 04 May 17:53
v1.12.0-rc2

We are pleased to release Cilium v1.12.0-rc2. The summary of changes are using as a base the version v1.12.0-rc1.

Summary of Changes

Major Changes:

  • Promote egress gateway to stable (#19320, @jibi)
  • Support setting service backend states such as quarantine, maintenance so that these backends are not selected for load-balancing service traffic. (#18814, @aditighag)

Minor Changes:

  • daemon: Split --bpf-lb-map-max into multiple options (#19326, @koncha99)
  • datapath: Allow egress GW with XDP (#19587, @brb)
  • egressgw: emit a warning rather than a fatal error when L7 proxy is enabled (#19608, @jibi)
  • helm: Create cilium IngressClass (#19524, @sayboras)
  • hubble/relay: Make the Hubble Peer service available by making it a Kubernetes service to eliminate the need to share a local Unix domain socket between a privileged pod (cilium daemon) and an unprivileged one (hubble-relay). (#18620, @nathanjsweet)
  • ingress: Add SocketOptions configuration (#19549, @sayboras)
  • ingress: Avoid plain text TLS secret in CEC (#19410, @sayboras)
  • ingress: Fix conformance tests for host-rules and path-rule (#19321, @sayboras)
  • ingress: Set max stream duration as 0 (#19550, @sayboras)
  • k8s: keep KVStore CiliumNode labels synced with Node object (#19375, @jibi)
  • Making operator aware of pending pod backlog on nodes for IP allocations (#19007, @hemanthmalla)
  • Remove privileged mode in Cilium's DaemonSet (#14446, @aanm)
  • vtep: VTEP map implementation to improve VTEP integration feature (#18824, @vincentmli)

Bugfixes:

  • bgpv1: Use IP address used for peering as a nexthop (#19402, @YutaroHayakawa)
  • clustermesh-apiserver: fixed nil pointer dereference (#18957, @abocim)
  • Consider VPC's secondary CIDRs during cilium_host IP restoration (#19341, @hemanthmalla)
  • Fatal when IPv6 is enabled but corresponding kernel modules are missing (#18941, @vadorovsky)
  • Fixed Cilium agent regression causing a crash due to ipcache controller being scheduled too soon. (#19501, @jrajahalme)
  • helm: Fix Hubble Service when ServiceMonitor is being used (#19220, @juissi-t)
  • Improve garbage collection for resources allocated by ToFQDNs policy for services which rotate IP addresses frequently such as Amazon S3 (#19452, @joestringer)
  • Only apply XDP acceleration for IPv6 Nodeport when enabled (with --bpf-lb-acceleration=native). (#19534, @julianwiedmann)
  • operator: fix identity GC collection (#19649, @aanm)
  • Revert Prometheus client to fix 'cilium metrics list' (#19496, @ti-mo)
  • Use identity labels for selector matching for Egress NAT Gateway (#19194, @blzhao-0)

CI Changes:

Misc Changes: