After upgrading to cdk v2, getting Stack name must be <= 128 characters

[Alon-Katz]

Describe the bug

I have been using cdk with version 1.147.1, and my cdk apps were successfully deployed.
After upgrading to cdk v2, I cant deploy my cdk app, getting "Stack name must be <= 128 characters" error message.

Looking at the stack name it tried to create, I cant understand what is going on, why is this name generated when I have provided a short logical id to all my resources and to the stack name itself.

I have a construct that exports an existing eks cluster (code is simplified just for example)

class EksCluster(Construct):
    def __init__(
        self,
        scope: Construct,
        stack_name: str,
        stage: str,
    ) -> None:
        super().__init__(scope, stack_name)
       self.cluster = self.import_cluster()

    def import_cluster(self) -> aws_eks.ICluster:
        return aws_eks.Cluster.from_cluster_attributes(
            self,
            id="imported-eks-cluster",
            cluster_name=cluster_name,
            kubectl_role_arn=kubectl_role_arn,
            open_id_connect_provider=open_id_provider,
            vpc=vpc,
        )

Then a simple app.py with the cdk app

app = App()
aws_lb_controller_stack = AwsLBControllerStack(
    app,
    f"aws-load-balancer-controller-{STAGE}",
    stage=STAGE,
    cluster_config=cluster_config,
    env=Environment(region=REGION, account=ACCOUNT),
)
app.synth()

And finally the stack itself, that uses the construct to use the eks cluster and create a service account for example

class AwsLBControllerStack(Stack):
    def __init__(
        self,
        scope: Construct,
        construct_id: str,
        stage: str,
        cluster_config: dict,
        **kwargs,
    ) -> None:
        super().__init__(scope, construct_id, **kwargs)
        self.eks_cluster = EksCluster(self, "eks-cluster", stage=stage)
        service_account = self.eks_cluster.cluster.add_service_account(
            "aws-lb-controller-service-account",
            name=AWS_LB_CONTROLLER_SERVICE_ACCOUNT_NAME,
            namespace=AWS_LB_CONTROLLER_NAMESPACE,
        )

This worked fine on cdk v1.147.1, but now after upgrading the cdk version and trying to deploy this code, it returns the following error:

jsii.errors.JSIIError: Stack name must be <= 128 characters. Stack name: 'awsloadbalancercontrollersandboxzolaawsloadbalancercontrollersandboxzolaeksclusterimportedeksekscluster980273B4KubectlProviderC92A5312'

Can you please explain what generated such a long name?
And most importantly, how can this be fixed?

Expected Behavior

Stack names should be short and understandable like they are defined in their logical ids, and most importantly it should allow me to deploy my cdk app without generating the name and then not allowing me to deploy when I did not define such name anywhere in the app.

Current Behavior

Stack names are generated with multiple duplicates inside and just unreadable names.

Reproduction Steps

Posted above a simple setup in python

Possible Solution

No response

Additional Information/Context

No response

CDK CLI Version

2.21.1

Framework Version

No response

Node.js Version

v17.9.0

OS

macOS monterey 12.3.1

Language

Python

Language Version

3.10.2

Other information

No response

[michaelsatish]

I ran into the problem stated here, and am unable to move forward in completing my CDK application. Just wondering if any timeline is established for this to get resolved?

[Alon-Katz]

The way I could solve this issue, was to shorten the base stack name to something unreadable (because I had to remove a lot of letters). This also created a downtime for our system because I had to delete the stack first and only then re-deploy with the shorten name.
This is unreasonable and if I knew it beforehand I would not upgrade to cdk v2 until this issue is resolved.

[michaelsatish]

@Alon-Katz Thank you for the comment. I can synthesize the stack with shorten name. No ideal, but works till this gets fixed.

[comcalvi]

@michaelsatish how is STAGE defined in your reproduction example? I haven't quite been able to reproduce this yet, but I have been able to synthesize a stack which contains nested stacks with long logical IDs, eg;

awsloadbalancercontrollerSTAGEeksclusterimportedekscluster8787A0B1KubectlProviderNestedStackawsloadbalancercontrollerSTAGEeksclusterimportedekscluster8787A0B1KubectlProviderNestedStackResourceE69C9315

This was present in the template produced by this app. cdk synth executes successfully. This isn't what you described though; could you share a full app that reproduces this exactly?

Stack code:

from aws_cdk import (
    core as cdk,
    aws_ec2 as ec2,
    aws_eks as eks,
)

from aws_cdk import core

class EksCluster(core.Construct):
    def __init__(
        self,
        scope: core.Construct,
        stack_name: str,
        stage: str,
    ) -> None:
        super().__init__(scope, stack_name)
        self.cluster = self.import_cluster()

    def import_cluster(self) -> eks.ICluster:
        return eks.Cluster.from_cluster_attributes(
            self,
            id="imported-eks-cluster",
            cluster_name="magic-cluster",
            kubectl_role_arn="arn:aws:iam::123456789012:role/MagicServiceRole",
            open_id_connect_provider=eks.OpenIdConnectProvider(self, 'openIdProvider', url='magic-url'),
            vpc=ec2.Vpc(self, "vpc"),
        )

class AwsLBControllerStack(core.Stack):
    def __init__(
        self,
        scope: core.Construct,
        construct_id: str,
        stage: str,
        #cluster_config: dict,
        **kwargs,
    ) -> None:
        super().__init__(scope, construct_id, **kwargs)
        clust = EksCluster(scope=self, stack_name="eks-cluster", stage=stage)
        service_account = clust.cluster.add_service_account(
            "aws-lb-controller-service-account",
        )

app.py:

from aws_cdk import core

from python_reprod.python_reprod_stack import AwsLBControllerStack

STAGE = "STAGE"

app = core.App()
aws_lb_controller_stack = AwsLBControllerStack(
    app,
    f"aws-load-balancer-controller-{STAGE}",
    stage=STAGE,
    #cluster_config=cluster_config,
    env=core.Environment(region='us-east-1', account="123456789012"),
)
app.synth()

[michaelsatish]

Hi @comcalvi,

Here is an example stack code to reproduce the error. Hope this helps.

package versions
aws-cdk-lib = "==2.25.0"
constructs = "==10.1.16"

app.py

#!/usr/bin/env python3
from aws_cdk import App

from pipeline import PipelineStack

app = App()
PipelineStack(app, "ApplicationPipelineStack")
app.synth()

pipeline.py

from aws_cdk import Stack
from aws_cdk import aws_s3 as s3
from aws_cdk import pipelines
from constructs import Construct

from stages import ApplicationStage


class PipelineStack(Stack):
    def __init__(self, scope: Construct, construct_id: str, **kwargs) -> None:
        super().__init__(scope, construct_id, **kwargs)

        bucket = s3.Bucket(self, "ApplicationCodeBucket")

        the_pipeline = pipelines.CodePipeline(
            self,
            "CdkPipeline",
            self_mutation=False,
            synth=pipelines.ShellStep(
                id="Synth",
                input=pipelines.CodePipelineSource.s3(bucket, "code"),
                commands=[
                    "npm install -g aws-cdk@latest",
                    "python3 -m pip install --user pipenv",
                    "pipenv install --deploy --system",
                ],
            ),
        )

        application_stage = ApplicationStage(self, "ApplicationStage")
        the_pipeline.add_stage(application_stage)

stages.py

from aws_cdk import Stage
from constructs import Construct

from stacks import ApplicationStack


class ApplicationStage(Stage):
    def __init__(self, scope: Construct, construct_id: str, **kwargs) -> None:
        super().__init__(scope, construct_id, **kwargs)

        _ = ApplicationStack(self, "ApplicationStack")

stacks.py

from aws_cdk import Stack
from aws_cdk import aws_eks as eks
from aws_cdk import aws_iam as iam
from constructs import Construct


class ApplicationStack(Stack):
    def __init__(self, scope: Construct, construct_id: str, **kwargs) -> None:
        super().__init__(scope, construct_id, **kwargs)

        cluster = eks.Cluster.from_cluster_attributes(
            self,
            "ApplicationCluster",
            cluster_name="ApplicationDevCluster",
            kubectl_role_arn="arn:aws:iam::12345:role/service-role/AmazonExample",
            open_id_connect_provider=iam.OpenIdConnectProvider(
                self, "ApplicationClusterIdentityProvider", url="http://"
            ),
        )

        service_acount = cluster.add_service_account("ApplicationServiceAccount")

        cluster.add_helm_chart(
            "ApplicationHelmChart",
            chart="nginx-ingress",
            repository="https://helm.nginx.com/stable",
            namespace="kube-system",
            values={"service_acount": service_acount.service_account_name},
        )

cdk ls

jsii.errors.JSIIError: Stack name must be <= 128 characters. Stack name: 'ApplicationStage-ApplicationStackApplicationPipelineStackApplicationStageApplicationStackApplicationClusterF3918940KubectlProvider5C2B31FB'

[comcalvi]

Thanks @michaelsatish, that reproduces the issue. I'll be working on a fix for this asap.

[github-actions]

⚠️COMMENT VISIBILITY WARNING⚠️

Comments on closed issues are hard for our team to see.
If you need more assistance, please either tag a team member or open a new issue that references this one.
If you wish to keep having a conversation with other community members under this issue feel free to do so.