antrea-io/antrea v2.1.0

Antrea v2.1.0

on GitHub

Added

• Add a new feature BGPPolicy that allows users to run a BGP process on selected Kubernetes Nodes and advertise Service IPs, Pod IPs, and Egress IPs to remote BGP peers.
  • Add BGPPolicy API and Controller. (#6009 #6203, @hongliangl)
  • Add BGP datapath interface and implement goBGP integration. (#6447, @hongliangl)
  • Add documentation for the BGPPolicy feature. (#6524, @hongliangl)
  • Refer to this document for more information about this feature.
• Add a new feature NodeLatencyMonitor that allows users to do east/west connectivity monitoring and provides an API to query inter-Node latency. (#6120 #6392 #6479, @IRONICBo)
• Add two new antctl commands to validate a K8s Cluster before Antrea installation, and allow users to validate basic network and security functionalities after Antrea is installed.
  • Add antctl check cluster command to to ensure that a K8s cluster is configured properly before Antrea installation. (#6278, @kanha-gupta)
  • Add antctl check installation command to conduct Pod/Service connectivity checks and verify basic NetworkPolicy rules. (#6133 #6313 #6367, @kanha-gupta)
  • Add documentation for the new antctl check commands. (#6373, @kanha-gupta)
  • Refer to this document for more information about these new commands.

Changed

• Ensure Antrea Proxy handles all Service traffic with proxyAll enabled, even when kube-proxy is present. (#6308, @hongliangl)
• Optimize the containerized OVS installation on Windows, manual installation of the OVS kernel driver is usually not required anymore. (#6383, @wenyingd)
• Add OVS driver installation in initContainer for Antrea Agent on Windows. (#6312, @XinShuYang)
• Use HostProcess container base image and buildx to build the Antrea Agent Windows image, Window Server 2019 and later are now supported with the same image, including Windows Server 2022. (#6325, @wenyingd)
• Support shared LoadBalancerIP for multiple Services by introducing a new annotation service.antrea.io/allow-shared-load-balancer-ip: true. (#6480, @tnqn)
• Promote feature CleanupStaleUDPSvcConntrack from Alpha to Beta. (#6372, @hongliangl)
• Always include Pod labels in FlowAggregator IPFIX template. (#6418, @antoninbas)
• Fix live config updates on IPFIXExporter for FlowAggregator. (#6385, @antoninbas)
• Improve handling of config changes in FlowAggregator to support updating recordContents.podLabels at runtime. (#6378, @antoninbas)
• Add an EndpointResolver to remove Antrea Agent's dependency on proxy to access Antrea Service. (#6361, @antoninbas)
• Replace bincover with built-in Go coverage profiling tool. (#6090, @shikharish)
• Trim unneeded fields stored in informers and Node objects to reduce memory footprint. (#6317 #6351, @tnqn)
• Remove stale multicast routes to improve the readability of multicast routes. (#3242, @ceclinux)
• Add srcPodIP field in Traceflow observations. (#6247, @Atish-iaf)
• Use Helm to generate Antrea Windows manifests. (#6360, @shikharish)
• Upgrade CNI plugins from v1.4.1 to v1.5.1. (#6475, @antoninbas)
• Add documentation for the sameLabels feature in Antrea ClusterNetworkPolicy. (#6280, @Dyanngg)
• Add recommended usage of FQDN policies. (#6389, @Dyanngg)

Fixed

• Fix NodePortLocal rules being deleted incorrectly due to PodIP recycle. (#6531, @tnqn)
• Fix "Access is denied" error when importing certificates into the trusted publishers store on Windows server 2022. (#6529, @wenyingd)
• Fix the Node network connection breaking when attaching a host interface to the secondary OVS bridge. (#6504, @wenyingd)
• Exclude terminated Pods from group members when calculating AppliedToGroups and EgressGroups to prevent NetworkPolicies or Egresses applying to wrong Pods. (#6508, @tnqn)
• Fix install_cni_chaining script not configuring CNI conf correctly with AKS or CNI chaining, when the CNI conf file is not ready. (#6506, @tnqn)
• Wait for OVS bridge datapath ID to be available after creating br-int to avoid failures when the Antrea Agent tries to query supported OVS datapath features. (#6472, @antoninbas)
• Fix a bug causing Antrea Proxy not to delete stale UDP conntrack entries for the virtual NodePort DNAT IP. (#6379, @hongliangl)
• Fix Antrea Agent crash when enabling proxyAll in networkPolicyOnly mode. (#6259, @hongliangl)
• Improve stale UDP conntrack entries deletion accuracy in Antrea Proxy. (#6193, @hongliangl)
• Remove unexpected altname after renaming interface to avoid failure when moving host interface to OVS bridge. (#6321, @gran-vmv)
• Avoid generating a zombie process when starting Suricata, the L7 ANP engine. (#6366, @hongliangl)
• Fix a single rule deletion bug for NodePortLocal on Linux and improve robustness of NPL rule cleanup. (#6284, @antoninbas)
• Delay removal of flow-restore-wait to fix traffic interruption issue when Antrea Agent restarts. (#6342, @antoninbas)
• Fix antctl mc deploy command usage to make the version parameter optional. (#6287, @roopeshsn)
• Fix inaccuracy in Traceflow user guide. (#6319, @antoninbas)