Releases: aws/amazon-vpc-cni-k8s
v1.6.3 Release
The most important change in this release is a fix for #1008, the CNI pod crashloopbackoff due to incorrect instance metadata while fetching ENI issue, fixed in PR #1011.
Note! If you edit an existing v1.5.x CNI deployment, make sure to mount /var/run/dockershim.sock correctly!
Changes since v1.6.2
- Bug - Handle stale instance metadata (#1011, @mogren)
- Improvement - Add support for c5a and c5ad (#1003, @mogren)
- Improvement - Make the aws-cni-support.sh executable (#1007, @jayanthvn)
If you want to apply this config to one of your clusters:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.3/config/v1.6/aws-k8s-cni.yaml
If you instead edit your current deployment, make sure to mount /var/run/dockershim.sock correctly!
Verify the update:
kubectl describe daemonset aws-node -n kube-system | grep Image | cut -d "/" -f 2
amazon-k8s-cni:v1.6.3
To use version v1.6.3 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.3/config/v1.6/cni-metrics-helper.yaml
v1.7.0 Release Candidate 2
Second release candidate for CNI v1.7.0. This pre-release image is only available in us-west-2.
Additional changes since v1.7.0-rc1:
- Improvement - Add ipamd changes for sg support (#1126, @mogren)
- Improvement - Add support to setup pod network using VLANss (#1125, @SaranBalaji90)
- Improvement - Improve CRI->checkpoint logic in the face of downgrades (#1123, @anguslees)
- Docs - Changed data type for variables in README (#1116, @abhinavmpandey08)
- Improvement - Slash and burn unused code (#1115, @anguslees)
- Improvement - Remove references to unused metadata
owner-id(#1111, @anguslees) - Improvement - Remove old pre-1.3 migration code (#1110, @anguslees)
- Testing - Create script to run all release tests (#1106, @bnapolitan)
- Improvement - Enable log config for the metrics agent (#1104, @mogren)
- Improvement - Refactor ENI limit struct (#1035, @mogren)
- Testing - Cover bottlerocket cluster test (#1096, @bnapolitan)
- Testing - Introduce automated performance testing (#1068, @bnapolitan)
Thanks to everyone who contributed to this release; @SaranBalaji90, @anguslees, @abhinavmpandey08 and @bnapolitan!
If you want to test this release candidate in one of your clusters, you can use this sample aws-node daemonset config:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.7/config/v1.7/aws-k8s-cni.yaml
Verify the update:
kubectl describe ds aws-node -n kube-system | grep Image | cut -d ":" -f 2-3
602401143452.dkr.ecr.us-west-2.amazonaws.com/amazon-k8s-cni-init:v1.7.0-rc2
602401143452.dkr.ecr.us-west-2.amazonaws.com/amazon-k8s-cni:v1.7.0-rc2
v1.6.4 Release
This is a patch release to address issue #711, Duplicate IP getting assigned during IPAMD restart. This was fixed in PR #1118.
Changes since v1.6.3
- Bug - Use docker/CRI to discover pods at node init(#1118 , @fawadkhaliq / @anguslees)
- Improvement - ARM64 support (through improved build setup and ECR manifest)
If you want to apply this config to one of your clusters:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.4/config/v1.6/aws-k8s-cni.yaml
Verify the update:
kubectl describe daemonset aws-node -n kube-system | grep Image | cut -d "/" -f 2
amazon-k8s-cni:v1.6.4
To use version v1.6.4 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.4/config/v1.6/cni-metrics-helper.yaml
v1.6.4 Release Candidate 1
First release candidate for CNI v1.6.4. This pre-release image is only available in us-west-2.
The main reason for this patch release to address issue #711 by cherry picking the fix in PR #972 to the 1.6 branch.
Bug - Use docker/CRI to discover pods at node init(#1118 , @fawadkhaliq / @anguslees )
If you want to test this release candidate in one of your clusters, you can use this sample aws-node daemonset config:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.4/config/v1.6/aws-k8s-cni.yaml
Verify the update:
kubectl describe ds aws-node -n kube-system | grep Image | cut -d ":" -f 2-3
602401143452.dkr.ecr.us-west-2.amazonaws.com/amazon-k8s-cni:v1.6.4-rc1
To test version v1.6.4-rc1 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.4/config/v1.6/cni-metrics-helper.yaml
v1.7.0 Release Candidate 1
First release candidate for CNI v1.7.0. This pre-release image is only available in us-west-2 and there are couple of changes that require some extra attention:
- #972 - We persist internal state of the
aws-nodepod to a file and use that to recover after a restart. Because of that, just changing the image tag will not work. Please update using the latest config file. - #955 - Added init container to copy portmap and loopback binaries, and set up
rp_filter. This means theaws-nodepermissions can be reduced fromprivilegedto just"NET-ADMIN"
Full changelog
- Bug - Use limits from API for g4dn.16xlarge (#1086, @mogren)
- Improvement - Use sed as a stream editor and redirect to file (#1069, @willejs)
- Docs - Fix docs links for cni-metrics-agent (#1072, @mogren)
- Bug - Make metrics-helper docker logging statement multi-arch compatible (#1067, @nprab428)
- Improvement - JSON output format for the entrypoint script (#1066, @jayanthvn)
- Testing - scripts/lib: bump up tester to v1.4.0 (#1065, @gyuho)
- Improvement - Use install command instead of cp (#1061, @mogren)
- Improvement - Updated manifest configs with default env vars (#1057, @saiteja313)
- Improvement - Default to random-fully (#1048, @mogren)
- Improvement - Update probe settings (#1028, @mogren)
- Improvement - Added warning if delete on termination is set to false for the primary ENI (#1024, @jayanthvn)
- Testing - Add parallel testing to conformance (#1018, @bnapolitan)
- Testing - Cache go packages in CircleCI (#1017, @bnapolitan)
- Testing - Create roles by default for e2e test cluster creation (#994, @bnapolitan)
- Improvement - Limit scope of logs writable by ipamd container (#987, @anguslees)
- Improvement - Autogenerate per-region YAML manifests from a common template (#986, @anguslees)
- Improvement - Persist IPAM state to local file and use across restarts (#972, @anguslees)
- Improvement - Add init container (#955, @mogren)
- Improvement - Refresh subnet/CIDR information periodically (#903, @nithu0115)
Thanks to everyone who contributed to this release, @anguslees, @bnapolitan, @gyuho, @jayanthvn, @nithu0115, @nprab428, @saiteja313, @willejs
If you want to test this release candidate in one of your clusters, you can use this sample aws-node daemonset config:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.7/config/v1.7/aws-k8s-cni.yaml
Verify the update:
kubectl describe ds aws-node -n kube-system | grep Image | cut -d ":" -f 2-3
602401143452.dkr.ecr.us-west-2.amazonaws.com/amazon-k8s-cni:v1.7.0-rc1
To test version v1.7.0-rc1 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.7/config/v1.7/cni-metrics-helper.yaml
v1.6.3 Release Candidate 1
First release candidate for CNI v1.6.3. This pre-release image is only available in us-west-2.
The main reason for this patch release to address issue #1008 by cherry picking the fix in PR #1011 to the 1.6 branch.
- Bug - Handle stale instance metadata (#1011, @mogren)
- Improvement - Add support for c5a and c5ad (#1003, @mogren)
- Improvement - Make
aws-cni-support.shexecutable (#1007, @jayanthvn)
If you want to test this release candidate in one of your clusters, you can use this sample aws-node daemonset config:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.3/config/v1.6/aws-k8s-cni.yaml
Verify the update:
kubectl describe ds aws-node -n kube-system | grep Image | cut -d ":" -f 2-3
602401143452.dkr.ecr.us-west-2.amazonaws.com/amazon-k8s-cni:v1.6.3-rc1
To test version v1.6.3-rc1 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.3/config/v1.6/cni-metrics-helper.yaml
v1.6.2 Release
The most important change in this release is a fix for #931, the gRPC Unix-socket proxy issue. Other changes include some additional debug information, improved integration tests, setting the user agent in EC2 requests, and updated limits for m6g, c6g and r6g instance types.
Note! If you edit an existing v1.5.x CNI deployment, make sure to mount /var/run/dockershim.sock correctly!
Thanks to everyone who contributed to this release: @gyuho, @mogren, @nithu0115 and @dthorsen
Changes since v1.6.1
- Improvement - Fix order of file copies in entrypoint.sh (#935, @dthorsen)
- Improvement - Check all errors and log appropriately (#939, @mogren)
- Improvement - Add MTU and RPFilter configs to debug (#954, @mogren)
- Improvement - Bump aws-k8s-tester to v1.2.2 (#978, @gyuho)
- Improvement - Add context and user agent to EC2 requests (#979, @mogren)
- Improvement - Update limits for m6g, c6g and r6g (#996, @mogren)
- Bug - Add WithNoProxy to ignore proxies in gRPC connections when using unix sockets (#980, @nithu0115)
If you want to apply this config to one of your clusters:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.2/config/v1.6/aws-k8s-cni.yaml
If you instead edit your current deployment, make sure to mount /var/run/dockershim.sock correctly!
Verify the update:
kubectl describe daemonset aws-node -n kube-system | grep Image | cut -d "/" -f 2
amazon-k8s-cni:v1.6.2
To use version v1.6.2 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.2/config/v1.6/cni-metrics-helper.yaml
v1.6.2 Release Candidate 1
First release candidate for CNI v1.6.2. This pre-release image is only available in us-west-2.
This is a patch release to address the issue #931. The fix for this was added in PR #980 by @nithu0115.
If you want to test this release candidate in one of your clusters you can use this aws-node daemonset config:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.2/config/v1.6/aws-k8s-cni.yaml
Verify the update:
kubectl describe daemonset aws-node -n kube-system | grep Image | cut -d "/" -f 2
amazon-k8s-cni:v1.6.2-rc1
To test version v1.6.2 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.2/config/v1.6/cni-metrics-helper.yaml
v1.6.1 Release
Major improvements include a complete rewrite of the Makefile, switching to Zap for structured logging, a big reduction in number of DescribeNetworkInterfaces calls to EC2 for large clusters and improved integration tests.
Note! If you edit an existing v1.5.x CNI deployment, make sure to mount /var/run/dockershim.sock correctly!
Thanks to everyone who contributed to this release: @anguslees, @groodt, @gyuho, @jahkeup, @jaypipes, @lmm, @marcincuber, @nithu0115, @ricardochimal and @SaranBalaji90
Changes since v1.6.0
- Feature - Support architecture targeted builds (#837, @jahkeup)
- Feature - Zap logger (#824, @nithu0115)
- Improvement - Run conformance test as part of PR/Release certification (#851, @SaranBalaji90)
- Improvement - Use eks:cluster-name as clusterId (#856, @groodt)
- Improvement - Bump Calico to v3.13.0 (#857, @lmm)
- Improvement - Use go.mod version of mockgen (#863, @anguslees)
- Improvement - Mock /proc/sys (#870, @anguslees)
- Improvement - Replace debug script with updated script from EKS AMI (#864, @mogren)
- Improvement - Update cluster-proportional-autoscaler to 1.7.1 (#885, @ricardochimal)
- Improvement - Remove unnecessary/incorrect ClusterRole resource (#883, @anguslees)
- Improvement - Disable IPv6 RA and ICMP redirects (#897, @anguslees)
- Improvement - scripts/lib/aws.sh: use "aws-k8s-tester" v1.0.0 (#900, @gyuho)
- Improvement - Configure rp_filter based on env variable (#902, @SaranBalaji90)
- Improvement - Less verbose logging (#908, @mogren)
- Improvement - Reduce number of calls to EC2 API (#909, @mogren)
- Improvement - Bump containernetworking dependencies (#916, @mogren)
- Improvement - Use -buildmode=pie for binaries (#919, @mogren)
- Bug - Add missing permissions in typha-cpha sa (Calico) (#892, @marcincuber)
- Bug - Fix logging to stdout (#904, @mogren)
- Bug - Ensure non-nil Attachment in getENIAttachmentID (#915, @jaypipes)
If you want to apply this config to one of your clusters:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.1/config/v1.6/aws-k8s-cni.yaml
If you instead edit your current deployment, make sure to mount /var/run/dockershim.sock correctly!
Verify the update:
kubectl describe daemonset aws-node -n kube-system | grep Image | cut -d "/" -f 2
amazon-k8s-cni:v1.6.1
To use version v1.6.1 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.1/config/v1.6/cni-metrics-helper.yaml
v1.6.1 Release Candidate 1
First release candidate for CNI v1.6.1. Major improvements include a complete rewrite of the Makefile, switching to Zap for structured logging, a big reduction in number of DescribeNetworkInterfaces calls to EC2 for large clusters and improved integration tests.
If you want to test this release candidate in one of your clusters you can use this aws-node daemonset config:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.1/config/v1.6/aws-k8s-cni.yaml
If you instead edit your current deployment, make sure to mount /var/run/dockershim.sock correctly!
Verify the update:
kubectl describe daemonset aws-node -n kube-system | grep Image | cut -d "/" -f 2
amazon-k8s-cni:v1.6.1-rc1
To test version v1.6.1 of the cni-metrics-helper in a cluster:
kubectl apply -f https://raw.githubusercontent.com/aws/amazon-vpc-cni-k8s/release-1.6.1/config/v1.6/cni-metrics-helper.yaml