Weekly Status: Week 36 2021

[cwayne18]

Weekly Report

Weekly status report for k3s-io/k3s Week #36 2021

Here's what the team has focused on this week:

• Ongoing 1.22 work (RC available here)
• Cross-team work with Rancher team
  • Continued work on enabling dual-stack (1.22)
  • In flannel integration in k3s
  • In kube-router (still in progress)
• Created k3s-engine branch as base for downstream projects that are ahead of release-1.21, but not ready for the 1.22 changes on master
• Created sonobouy plugin to run integration tests against existing clusters
• Add support for stargz snapshotter (1.22)
• Add support for migration from sqlite to etcd (1.22)
• Add support for etcd backups to insecure S3 endpoints (local minio, for example) (1.22)
• Add support for kine database size reported via apiserver etcd_db_total_size_in_bytes metric (1.22)
• Updated cni-plugins version (1.22)
• On-going focus on process improvement/automation

Weekly Stats

	Opened this week	Closed this week
Issues	14	25
PR's	8	8

New stars	174
New forks	13

PR's Closed

• #3989: [Release-1.19] Backport cluster-reset fix
• #3987: [engine-1.21] Add exposed metrics listener instead of replacing loopback listener
• #3986: [release-1.21] Add exposed metrics listener instead of replacing loopback listener
• #3985: Add exposed metrics listener instead of replacing loopback listener
• #3978: SupportPodPidsLimit is locked to true of 1.20, making pids cgroup support mandatory
• #3975: Properly handle fatal errors and context cancelation
• #3968: Allow option to disable s3 over https when using etcd-snapshot
• #3231: Migrate sqlite data to etcd when initializing the cluster

PR's Opened

• #3989: [Release-1.19] Backport cluster-reset fix
• #3987: [engine-1.21] Add exposed metrics listener instead of replacing loopback listener
• #3986: [release-1.21] Add exposed metrics listener instead of replacing loopback listener
• #3985: Add exposed metrics listener instead of replacing loopback listener
• #3978: SupportPodPidsLimit is locked to true of 1.20, making pids cgroup support mandatory
• #3975: Properly handle fatal errors and context cancelation
• #3974: [dual-stack] Flannel unit tests
• #3968: Allow option to disable s3 over https when using etcd-snapshot

Issues Opened

• #3988: [Release-1.19] Passing cluster-reset fails to trigger reset due to force-new-cluster value set to false
• #3984: [release-1.21] etcd-expose-metrics setting replaces loopback listener with exposed listener
• #3983: etcd-expose-metrics setting replaces loopback listener with exposed listener
• #3982: Migrate sqlite data to etcd when initializing the cluster
• #3981: Can't pull the images after starting the k3s
• #3979: How to configure kubelet to adjust allocatable resources
• #3977: K3s exits with lengthy stack trace when kube-proxy cannot parse args
• #3976: Update documentation to reflect new memory requirements for newer kubernetes
• #3973: etcd member management controller
• #3972: Allow backing up etcd snapshots to work with an insecure s3 endpoint
• #3970: Unable to dual-stack k3s API server
• #3969: wrong permissions on /etc/rancher/k3s/k3s.yaml
• #3967: not effect to change cluster and service cidr
• #3966: build source with deepsource.io, report risk bug for pkg/agent/containerd/config_linux.go

Issues Closed

• #3979: How to configure kubelet to adjust allocatable resources
• #3969: wrong permissions on /etc/rancher/k3s/k3s.yaml
• #3945: [release-1.19] apiaddresses in etcd are being overwritten by pod ips scheduled to server nodes
• #3944: [release-1.20] apiaddresses in etcd are being overwritten by pod ips scheduled to server nodes
• #3943: [release-1.21] apiaddresses in etcd are being overwritten by pod ips scheduled to server nodes
• #3942: [master] apiaddresses in etcd are being overwritten by pod ips scheduled to server nodes
• #3933: Allow setting a custom node.kubernetes.io/instance-type node label
• #3927: Implement Sonobuoy Plugin for Integration tests
• #3876: [Release-1.19] Restoration fails on etcd only nodes due to invalid state of loadbalancers
• #3875: [Release-1.20] Restoration fails on etcd only nodes due to invalid state of loadbalancers
• #3874: [Release-1.21] Restoration fails on etcd only nodes due to invalid state of loadbalancers
• #3873: [master] Restoration fails on etcd only nodes due to invalid state of loadbalancers
• #3817: The waitManagementIP function in pkg/daemons/agent/agent_windows.go is calling the incorrect HNS function.
• #3811: s3 snapshots do not show in the rancher 2.6 UI
• #3787: Systemd does not get notified with READY state in etcd only nodes
• #3780: Cut 1.20.10
• #3745: snapshot creation without sudo should return a permission denied error message
• #3535: Networking broken on master node after deploying hostNetwork=true Pod.
• #3373: Update embedded kube-router network policy controller
• #3258: k3s generated iptables input filter rules are not correct
• #2977: K3s deployed on CentOS8 cannot join rancher
• #2971: ProgressDeadlineExceeded is not fired
• #2963: If external IP is defined K3s API NAT rule use it.
• #2509: Master node wont rejoin etcd cluster after reboot
• #66: Plans for ZFS support?

Community PRs Closed

• #3968: Allow option to disable s3 over https when using etcd-snapshot

Community PRs Opened

• #3968: Allow option to disable s3 over https when using etcd-snapshot