- Support for the ENI_CONFIG_LABEL_DEF environment variable #411
- feat(nodegroup): Support encryption of the root block device for nodes #407
- fix(ex/default-sg): rm Output tag values per string type reqs #404
- nodegroup(asgName): fix asgName definition #401
- Cutting new release to include missing generated API docs from v0.19.1
- feat(nodegroup): add opt to attach extra security groups #390
- feat(cluster): add encryptionConfigKeyArn opt to encrypt k8s Secrets #389
For a more detailed list of the changes introduced in this release, please visit #381.
- fix(dashboard): disable dashboard from deploying if not set #378
- fix(cluster): use scoped kubeconfig with non-default AWS credentials #367
- Update node & go pulumi deps to 2.0 #375
- fix(aws): rm sync invokes for AWS data source calls #373
- refactor(aws-auth): replace aws-iam-authenticator with aws eks get-token
#362
- Note: for existing clusters, this change will recompute the kubeconfig
used, as its auth arguments and settings get updated to work with
aws eks get-token
. It should not affect cluster access or cause replacements of existing k8s resources.
- Note: for existing clusters, this change will recompute the kubeconfig
used, as its auth arguments and settings get updated to work with
- feat(nodegroup): use the latest recommended AMIs from the SSM store #366
- feat(cluster): support HTTP(S) proxy for cluster readiness & OIDC config #365
- deps(pulumi): bump node and go pulumi/pulumi to v1.13.1 #361
- feat(cluster): add getKubeconfig method to generate scoped kubeconfigs #356
- fix(oidc): Fix issue in OIDC getThumbprint helper function #346
- fix(oidc): use thumbprint of the intermediate root CA #342
- update(cni): update from v1.5.3 -> v1.6.0 #325
- fix(storageClasses): fix userStorageClass initialization #336
- feat(cluster): allow optional configuration of cluster name #322
- feat(identity): add support to setup OIDC provider #320
- Refactor managed nodegroup API and require its role be provided to the cluster #302
- Update pulumi/pulumi and re-enable withUpdate tests #327
- Fix js-yaml dependency changes in pulumi/k8s #324
- Unblock CI by disabling debug logging, rm unnecessary tests, and fixing broken tests #309
- feat(cluster): Support public access controls #295
- feat(cluster): Add cluster tagging #262
- refactor(vpcCni): set node anti-affinity to not deploy to fargate #291
- build: Upgrade to go1.13.4 #290
- feat(nodes): add support for Fargate #283
- feat(nodes): add createManagedNodeGroup #280
- fix(vpc-cni): allow logLevel & logFile to be set, or defaulted if not #274
- Update pulumi to 1.4.0 #270
- refactor(cluster): allow ClusterOptions to accept NodeGroupOptions #259
- Add new publicSubnetIds and privateSubnetIds cluster options. Also, update tests to use new awsx.ec2.Vpc API and new subnet options #238
- fix(iam): improve YAML error handling & reporting in IAM ops #231
- feat(iam): create eks cluster & resources with iam role provider #205
- fix(cni): read CNI YAML outside of the dynamic provider and update to v1.5.3 #223
- Revert "fix(cni): modify CNI filepath to store the relative path" #220
- Fix and improve migrate-nodegroup test (bump CNI from
v1.5.0
->v1.5.2
) #214 - fix(asgName): check 'NodeGroup' CFStack output key exists #213
- chore(cluster): add deprecation for kube-dashboard, customInstanceRolePolicy #202
- feat(storage-classes): export all user created storage classes #172
- update(eks): add example of migrating node groups with zero downtime #195
- refactor(secgroup): export createNodeGroupSecurityGroup & consolidate rules #183
- wait for EKS cluster endpoint to be available #193
- fix(cluster): support configuring private and public endpoint access #154
- fix(cluster): support passing additional arguments to /etc/eks/bootstrap.sh and --kubelet-extra-args #181
-
Default to a node AMI that matches the cluster version #175
-
fix(tags): rm ASG tag dupes, and consider tag inheritance for all tags #162
-
fix(nodegroup): make VPN-only subnets private #163
-
feature(cluster): Allow service role and instance profile to be injected during cluster creation #159
- ci(aws-iam-authenticator): use official S3 bucket to install bin #166
- fix(tags): change map types used in all tags to pulumi.Inputs of the map #157
- fix(cluster): expose instanceRoles #155
- tests(cluster): enable test to replace cluster by adding more subnets #150
- update(aws-k8s-cni): move from 1.4.1 -> 1.5.0 #148
- fix(cluster): rm dupe default storage class #136
- Expand nodejs SDK tests coverage, and add Kubernetes Smoke Tests for examples & tests #130
- update(aws-k8s-cni): move from 1.3.0 -> 1.4.1 #134
- fix(cluster): export missing instanceRoles in the cluster's CoreData #133
- fix(nodeSecurityGroupTags): only expose option through Cluster class #126
- fix(secgroups): do not null out ingress & egress
#128
- Note: This PR reverses the default null values used for the
ingress and egress in-line rules of the secgroups, introduced in
v0.18.3
. The null default was required to move to standalone secgroup rules, but it has introduced issues, and thus is being removed in this PR. - Upgrade Path - This is a breaking change unless you do the following steps:
- If using >=
v0.18.3
: update using the typical package update path. - If using <=
v0.18.2
:- First, update your cluster from using your current version to
v0.18.4
. - Next, update your cluster from
v0.18.4
tov0.18.5
(or higher) using the typical package update path.
- First, update your cluster from using your current version to
- If using >=
- Note: This PR reverses the default null values used for the
ingress and egress in-line rules of the secgroups, introduced in
- feat(tags): Set default tags & add opts: tags, and other resource tags #122
- feat(control plane logging): Enable control plane logging to cloudwatch. #100.
- fix(ami): only apply AMI smart-default selection on creation #114
- fix(secgroups): use standalone secgroup rules instead of in-line rules #109. Note, because we are replacing existing in-line secgroup rules with standalone rules, there may be a brief period of outage where the security group rules are removed before they get added back. This update happens in a matter of seconds (~5 sec), so any interruptions are short-lived.
- fix(nodegroup): filter on x86_64 arch for node AMI #112
- feat(nodePools): support per-nodegroup IAM instance roles #98
- Moves to the new 0.18.0 version of
@pulumi/aws
. Version 0.18.0 ofpulumi-aws
is now based on v2.2.0 of the AWS Terraform Provider, which has a variety of breaking changes from the previous version. See documentation in@pulumi/aws
repo for more details.
- Fix a bug where the regex used to retrieve Worker Node AMIs was not returning correct AMIs when either: specifying the master / control plane version, or relying on smart defaults of the lastest available image. #92
- feat(workers): add 'nodeAssociatePublicIpAddress' to toggle public IPs #81
- fix(getAmi): allow setting master version & explicitly filter Linux AMIs
#85
- Fix a bug where the wrong AMI was being returned due to a loosely defined regex.
- Add support for setting the master / control plane version of the cluster.
- Re-cut 0.17.1 as 0.17.2, due to a broken master branch caused by a pushed tag publishing the NPM package before master was able to.
- Support for
taints
onNodeGroups
. #63
- Depend on latest version of
@pulumi/pulumi
to get more precise delete before create semantics #46
- Expose the AutoScalingGroup on NodeGroups. #53
- Fix a bug where
desiredCapacity
was not being handled correctly. #55
- Support for multiple Worker
NodeGroup
s connected to a single EKS cluster. #39 - Support for Spot instances in
NodeGroup
s. #49 - Support for adding cutom policies to node
InstanceRole
. #49 - Support for adding labels to each instance in a
NodeGroup
. #49
- Allow configuring the subnets that worker nodes use.
- Improve detection of public vs. private subnets.