Releases: poseidon/typhoon
Releases · poseidon/typhoon
v1.26.3
v1.26.2
v1.26.1
v1.26.0
- Kubernetes v1.26.0
- Update etcd from v3.5.5 to v3.5.6
- Update Cilium from v1.12.3 to v1.12.4
- Update flannel from v0.15.1 to v0.20.2
- Reminder: Modules are no longer published to the Terraform Module Registry (#1282)
AWS
- Migrate AWS launch configurations to launch templates (#1275)
- Starting Dec 31, 2022 AWS won't add new instance types/families to launch configurations
Addons
v1.25.4
- Kubernetes v1.25.4
- Update Calico from v3.24.1 to v3.24.5
- Allow Kubelet kubeconfig to drain nodes, if desired (#330)
- Re-enable Kubelet Graceful Node Shutdown (#1261)
- Introduce companion project poseidon/scuttle
- Link to new Mastodon account for release announcements
- Deprecate publishing to the Terraform Module Registry
- Typhoon docs have always shown using Git-based module sources, not the Terraform Module Registry
- Module usage should be
source = "git::https://github.com/poseidon/typhoon/...notsource = poseidon/kubernetes/... - Terraform's Module Registry requires subtree mirroring typhoon to special terraform-platform-kubernetes repos, only supports release versions (no commit SHAs or forks), only ever contained Flatcar Linux modules (not Fedora CoreOS) for historical reasons
- Note, this does not affect Terraform Providers like
poseidon/matchboxorposeidon/ct, the registry works well for providers
Fedora CoreOS
- Remove unused
Wants=network.targetfrometcd-member.service(#1254)
Cloud
- Remove defunct
delete-node.servicefrom worker node configurations (#1256)
Addons
v1.25.3
- Kubernetes v1.25.3
- Switch Kubernetes registry from
k8s.gcr.iotoregistry.k8s.iofor addons (#1246) - Update Cilium from v1.12.2 to v1.12.3 (#1253)
Azure
- Change default Azure
worker_typefromStandard_DS1_v2toStandard_D2as_v5(#1248)- Get 2 VCPU, 7 GiB, 12500Mbps (vs 1 VCPU, 3.5GiB, 750 Mbps)
- Small increase in pay-as-you-go price ($53.29 -> $62.78)
- Small increase in spot price ($5.64/mo -> $7.37/mo)
- Change from Intel to AMD EPYC (
D2as_v5cheaper thanD2s_v5)
Flatcar Linux
- Add Flatcar Linux ARM64 support on Azure (docs, #1251)
- Switch from Azure Hypervisor gen1 to gen2 (action required) (#1248)
- Run
az vm image terms accept --publish kinvolk --offer flatcar-container-linux-free --plan stable-gen2
- Run
Docs
- Remove old docs note about not supporting ARM64 with Calico
- Typhoon supports ARM64 with
cilium,calico, andflannel
- Typhoon supports ARM64 with
Addons
v1.25.1
- Kubernetes v1.25.1
- Update etcd from v3.5.4 to v3.5.5
- Update Cilium from v1.12.1 to v1.12.2
- Update Calico from v3.23.3 to v3.24.1
- Revert Kubelet Graceful Node Shutdown on worker nodes (#1227)
- Fix issue where non-critical pods are left in Error/Completed state on node shutdown
- Remove feature flag disable workaround for kubernetes/kubernetes#112081
- Kubernetes reverted
LocalStorageCapacityIsolationFSQuotaMonitoringback to alpha
- Kubernetes reverted
- Remove workaround for preventing
search .propagation in kubernetes/kubernetes#112135- Upstream Kubernetes fix
Addons
v1.25.0
- Kubernetes v1.25.0
- Disable LocalStorageCapacityIsolationFSQuotaMonitoring feature gate (#1220, fixes kubernetes#112081)
- Add workaround to revert adding "search ." to containers'
/etc/resolv.conf(#1224, fixes kubernetes#112135)
- Migrate most Kubelet flags to KubeletConfiguration file (#1219)
- Configure Kubelet Graceful Node Shutdown (#1222)
- Allow up to 30s for critical pods to gracefully shutdown on node shutdown
- Allow up to 15s for regular pods to gracefully shutdown on node shutdown
- Mark node NotReady promptly on node shutdown
- Lengthen systemd inhibitor lock max delay from 5s to 45s
Fedora CoreOS
- Change Podman
log-driverfromjournaldtok8s-file(#1221)- Fix
etcd-memberand Kubelet systemd service log lines appearing twice in journal logs
- Fix
v1.24.4
- Kubernetes v1.24.4
- Update CoreDNS from v1.8.6 to v1.9.3
- Update Cilium from v1.11.7 to v1.12.1
- Update Calico from v3.23.1 to v3.23.3
- Switch Kubernetes registry from
k8s.gcr.iotoregistry.k8s.io(#1206) - Remove use of deprecated Terraform template provider (#1194)
Flatcar Linux
- Migrate Flatcar Linux from Ignition spec v2.3.0 to v3.3.0 (#1196) (action required)
- Please modify any Flatcar Linux snippets to use the Butane Config format (action required)
variant: flatcar
version: 1.0.0
...Fedora CoreOS
- Remove ineffective
/etc/fedora-coreos/iptables-legacy.stamp(#1201)- Typhoon already uses iptables v1.8.7 (nf_tables) since FCOS 36
- Staying on legacy iptables required a file in
/etc/coreosinstead
AWS
- Refresh instances in autoscaling group when launch configuration changes (#1208) (docs, important)
- Worker launch configuration changes start an autoscaling group instance refresh to replace instances
- Instance refresh creates surge instances, waits for a warm-up period, then deletes old instances
- Changing
worker_type,disk_*,worker_price,worker_target_groups, or Butaneworker_snippetson existing worker nodes will replace instances - New AMIs or changing
os_streamwill be ignored, to allow Fedora CoreOS or Flatcar Linux to keep themselves updated - Previously, new launch configurations were made in the same way, but not applied to instances unless manually replaced
- Rename worker autoscaling group
${cluster_name}-worker(#1202)- Rename launch configuration
${cluster_name}-workerinstead of a random id
- Rename launch configuration
- Roll instance template changes to worker managed instance groups (#1207) (docs, important)
- Worker instance template changes roll out by gradually replacing instances
- Automatic rollouts create surge instances, wait for health checks, then delete old instances (0 unavailable instances)
- Changing
worker_type,disk_size,worker_preemptible, or Butaneworker_snippetson existing worker nodes will replace instances - New compute images or changing
os_streamwill be ignored, to allow Fedora CoreOS or Flatcar Linux to keep themselves updated - Previously, new instance templates were made in the same way, but not applied to instances unless manually replaced
- Add health checks to worker managed instance groups (i.e. "autohealing") (#1207)
- Use health checks to probe kube-proxy every 30s
- Replace worker nodes that fail the health check 6 times (3min)
- Name
kube-apiserverandworkerhealth checks consistently (#1207)- Use name
${cluster_name}-apiserver-healthand${cluster_name}-worker-health
- Use name
- Rename managed instance group from
${cluster_name}-worker-groupto${cluster_name}-worker(#1207) - Fix bug provisioning clusters with multiple controller nodes (#1195)