CodeStorageExceededException

What Does Code Storage Exceeded Mean?

CodeStorageExceededException is a hard ceiling on your account's "deployment debt." Every time you publish a new Lambda version or update a layer, AWS stores that package permanently. In CI/CD pipelines that deploy daily, these 50MB-100MB packages accumulate rapidly. Since the 75GB limit is shared across all functions in a region, one "noisy" function with hundreds of versions can block deployments for your entire infrastructure.

Common Causes

• -Unbounded Versioning: CI/CD pipelines calling PublishVersion: true on every push without deleting older, unused versions.
• -Orphaned Layers: Large dependency layers (e.g., Pandas, Scikit-learn) being updated frequently, with each 250MB version staying in storage indefinitely.
• -Large Deployment Packages: Using heavy container images or zips with unnecessary dev-dependencies that inflate the total storage footprint.
• -High Function Density: Having hundreds of micro-functions in a single account/region, each contributing to the 75GB total.

How to Fix Code Storage Exceeded

1. 1Delete Old Versions: Use a script to delete all function versions except the last 3-5 and those currently pointed to by an Alias.
2. 2Cleanup Unused Layers: Audit your Lambda Layers and delete versions that are no longer referenced by any Active function.
3. 3Request Quota Increase: If your workload genuinely requires >75GB, request an increase via AWS Service Quotas (though cleanup is usually the better first step).
4. 4Disable Auto-Publish: Set PublishVersion to false in your SAM/CDK/Serverless templates if you don't use Aliases for traffic shifting.

Step-by-Step Diagnosis for Code Storage Exceeded

1. 1Run aws lambda get-account-settings: This provides the TotalCodeSize and confirms if you are at the limit.
2. 2Identify the "Storage Hogs": List all functions and sort them by CodeSize * NumberOfVersions.
3. 3Check Aliases: Before deleting, ensure the version is not being used by a prod or staging alias.
4. 4Audit Layers: Check for old layer versions that are 0.5GB+ and haven't been updated in months.

aws lambda get-account-settings \
    --query 'AccountLimit.TotalCodeSize'

// Logic: List versions -> Sort by ID -> Keep top 5 -> Delete rest
const versions = await lambda.listVersionsByFunction({ FunctionName: "my-func" }).promise();
const toDelete = versions.Versions
  .filter(v => v.Version !== '$LATEST')
  .slice(0, -5); // Keep the newest 5

for (const v of toDelete) {
  await lambda.deleteFunction({ FunctionName: "my-func", Qualifier: v.Version }).promise();
}

How to Verify the Fix

• -Run get-account-settings again and confirm TotalCodeSize has significantly dropped.
• -Attempt a new deployment; it should now succeed without the CodeStorageExceeded error.
• -Verify that rollback capabilities (if needed) are still intact for the last few versions.

How to Prevent Recurrence

• -Automated Cleanup: Add a "post-deploy" step in your pipeline that runs a cleanup script to keep only the N most recent versions.
• -CloudWatch Alarms: Set an alarm on the AccountCodeSize metric to trigger at 80% (60GB) capacity.
• -Optimized Zips: Use .lambdaignore or similar tools to strip node_modules of dev-dependencies and binary assets.
• -Pro-tip: If you use container-image-based Lambdas, their storage is managed differently (ECR), but they still count toward specific account quotas. For zip-based Lambdas, aggressive version pruning is the only long-term solution.

Related Errors

Decision Support

Official References

• -Lambda: Managing Versions
• -Lambda: Deployment Quotas
• -AWS Official Documentation