Troubleshooting CodePipeline
The following information might help you troubleshoot common issues in AWS CodePipeline.
Topics
- Pipeline error: A pipeline configured with AWS Elastic Beanstalk returns an error message: "Deployment failed. The provided role does not have sufficient permissions: Service:AmazonElasticLoadBalancing"
- Deployment error: A pipeline configured with an AWS Elastic Beanstalk deploy action hangs instead of failing if the "DescribeEvents" permission is missing
- Pipeline error: A source action returns the insufficient permissions message: "Could not access the CodeCommit repository repository-name. Make sure that the pipeline IAM role has sufficient permissions to access the repository."
- Pipeline error: A Jenkins build or test action runs for a long time and then fails due to lack of credentials or permissions
- Pipeline error: A pipeline created in one AWS Region using a bucket created in another AWS Region returns an "InternalError" with the code "JobFailed"
- Deployment error: A ZIP file that contains a WAR file is deployed successfully to AWS Elastic Beanstalk, but the application URL reports a 404 not found error
- Pipeline artifact folder names appear to be truncated
- Add CodeBuild GitClone permissions for connections to Bitbucket, GitHub, GitHub Enterprise Server, or GitLab.com
- Add CodeBuild GitClone permissions for CodeCommit source actions
- Pipeline error: A deployment with the CodeDeployToECS action returns an error message: "Exception while trying to read the task definition artifact file from: <source artifact name>"
- GitHub (via OAuth app) source action: Repository list shows different repositories
- GitHub (via GitHub App) source action: Unable to complete the connection for a repository
- Amazon S3 error: CodePipeline service role <ARN> is getting S3 access denied for the S3 bucket <BucketName>
- Pipelines with an Amazon S3, Amazon ECR, or CodeCommit source no longer start automatically
- Connections error when connecting to GitHub: "A problem occurred, make sure cookies are enabled in your browser" or "An organization owner must install the GitHub app"
- Pipelines with execution mode changed to QUEUED or PARALLEL mode fails when run limit reached
- Pipelines in PARALLEL mode have an outdated pipeline definition if edited when changing to QUEUED or SUPERSEDED mode
- Pipelines changed from PARALLEL mode will display a previous execution mode
- Pipelines with connections that use trigger filtering by file paths might not start at branch creation
- Pipelines with connections that use trigger filtering by file paths might not start when file limit is reached
- CodeCommit or S3 source revisions in PARALLEL mode might not match EventBridge event
- Need help with a different issue?
Pipeline error: A pipeline configured with AWS Elastic Beanstalk returns an error message: "Deployment failed. The provided role does not have sufficient permissions: Service:AmazonElasticLoadBalancing"
Problem: The service role for CodePipeline does not have sufficient permissions for AWS Elastic Beanstalk, including, but not limited to, some operations in Elastic Load Balancing. The service role for CodePipeline was updated on August 6, 2015 to address this issue. Customers who created their service role before this date must modify the policy statement for their service role to add the required permissions.
Possible fixes: The easiest solution is to edit the policy statement for your service role as detailed in Add permissions to the CodePipeline service role.
After you apply the edited policy, follow the steps in Start a pipeline manually to manually rerun any pipelines that use Elastic Beanstalk.
Depending on your security needs, you can modify the permissions in other ways, too.
Deployment error: A pipeline configured with an AWS Elastic Beanstalk deploy action hangs instead of failing if the "DescribeEvents" permission is missing
Problem: The service role for CodePipeline must include the
"elasticbeanstalk:DescribeEvents"
action for any pipelines that use AWS Elastic Beanstalk.
Without this permission, AWS Elastic Beanstalk deploy actions hang without failing or indicating an error.
If this action is missing from your service role, then CodePipeline does not have permissions to run
the pipeline deployment stage in AWS Elastic Beanstalk on your behalf.
Possible fixes: Review your CodePipeline service role. If the
"elasticbeanstalk:DescribeEvents"
action is missing, use the steps in Add permissions to the CodePipeline
service role
to add it using the Edit Policy feature in the IAM console.
After you apply the edited policy, follow the steps in Start a pipeline manually to manually rerun any pipelines that use Elastic Beanstalk.
Pipeline error: A source action
returns the insufficient permissions message: "Could not access the CodeCommit repository
repository-name
. Make sure that the pipeline IAM role has sufficient
permissions to access the repository."
Problem: The service role for CodePipeline does not have sufficient permissions for CodeCommit and likely was created before support for using CodeCommit repositories was added on April 18, 2016. Customers who created their service role before this date must modify the policy statement for their service role to add the required permissions.
Possible fixes: Add the required permissions for CodeCommit to your CodePipeline service role's policy. For more information, see Add permissions to the CodePipeline service role.
Pipeline error: A Jenkins build or test action runs for a long time and then fails due to lack of credentials or permissions
Problem: If the Jenkins server is installed on an Amazon EC2 instance, the instance might not have been created with an instance role that has the permissions required for CodePipeline. If you are using an IAM user on a Jenkins server, an on-premises instance, or an Amazon EC2 instance created without the required IAM role, the IAM user either does not have the required permissions, or the Jenkins server cannot access those credentials through the profile configured on the server.
Possible fixes: Make sure that Amazon EC2 instance role or
IAM user is configured with the AWSCodePipelineCustomActionAccess
managed
policy or with the equivalent permissions. For more information, see AWS managed policies for AWS CodePipeline.
If you are using an IAM user, make sure the AWS profile configured on the instance uses the IAM user configured with the correct permissions. You might have to provide the IAM user credentials you configured for integration between Jenkins and CodePipeline directly into the Jenkins UI. This is not a recommended best practice. If you must do so, be sure the Jenkins server is secured and uses HTTPS instead of HTTP.
Pipeline error: A pipeline created in one AWS Region using a bucket created in another AWS Region returns an "InternalError" with the code "JobFailed"
Problem: The download of an artifact stored in an Amazon S3 bucket will fail if the pipeline and bucket are created in different AWS Regions.
Possible fixes: Make sure the Amazon S3 bucket where your artifact is stored is in the same AWS Region as the pipeline you have created.
Deployment error: A ZIP file that contains a WAR file is deployed successfully to AWS Elastic Beanstalk, but the application URL reports a 404 not found error
Problem: A WAR file is deployed successfully to an AWS Elastic Beanstalk environment, but the application URL returns a 404 Not Found error.
Possible fixes: AWS Elastic Beanstalk can unpack a ZIP file, but not
a WAR file contained in a ZIP file. Instead of specifying a WAR file in your
buildspec.yml
file, specify a folder that contains the content to be
deployed. For example:
version: 0.2 phases: post_build: commands: - mvn package - mv target/my-web-app ./ artifacts: files: - my-web-app/**/* discard-paths: yes
For an example, see AWS Elastic Beanstalk Sample for CodeBuild.
Pipeline artifact folder names appear to be truncated
Problem: When you view pipeline artifact names in CodePipeline, the names appear to be truncated. This can make the names appear to be similar or seem to no longer contain the entire pipeline name.
Explanation: CodePipeline truncates artifact names to ensure that the full Amazon S3 path does not exceed policy size limits when CodePipeline generates temporary credentials for job workers.
Even though the artifact name appears to be truncated, CodePipeline maps to the artifact bucket in a way that is not affected by artifacts with truncated names. The pipeline can function normally. This is not an issue with the folder or artifacts. There is a 100-character limit to pipeline names. Although the artifact folder name might appear to be shortened, it is still unique for your pipeline.
Add CodeBuild GitClone permissions for connections to Bitbucket, GitHub, GitHub Enterprise Server, or GitLab.com
When you use an AWS CodeStar connection in a source action and a CodeBuild action, there are two ways the input artifact can be passed to the build:
-
The default: The source action produces a zip file that contains the code that CodeBuild downloads.
-
Git clone: The source code can be directly downloaded to the build environment.
The Git clone mode allows you to interact with the source code as a working Git repository. To use this mode, you must grant your CodeBuild environment permissions to use the connection.
To add permissions to your CodeBuild service role policy, you create a customer-managed policy
that you attach to your CodeBuild service role. The following steps create a policy where the
UseConnection
permission is specified in the action
field, and the
connection ARN is specified in the Resource
field.
To use the console to add the UseConnection permissions
-
To find the connection ARN for your pipeline, open your pipeline and click the (i) icon on your source action. You add the connection ARN to your CodeBuild service role policy.
An example connection ARN is:
arn:aws:codeconnections:eu-central-1:123456789123:connection/sample-1908-4932-9ecc-2ddacee15095
-
To find your CodeBuild service role, open the build project used in your pipeline and navigate to the Build details tab.
-
Choose the Service role link. This opens the IAM console where you can add a new policy that grants access to your connection.
-
In the IAM console, choose Attach policies, and then choose Create policy.
Use the following sample policy template. Add your connection ARN in the
Resource
field, as shown in this example:{ "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": "codestar-connections:UseConnection", "Resource": "
insert connection ARN here
" } ] }On the JSON tab, paste your policy.
-
Choose Review policy. Enter a name for the policy (for example,
connection-permissions
), and then choose Create policy. -
Return to the page where you were attaching permissions, refresh the policy list, and select the policy you just created. Choose Attach policies.
Add CodeBuild GitClone permissions for CodeCommit source actions
When your pipeline has a CodeCommit source action, there are two ways you can pass the input artifact to the build:
-
Default – The source action produces a zip file that contains the code that CodeBuild downloads.
-
Full clone – The source code can be directly downloaded to the build environment.
The Full clone option allows you to interact with the source code as a working Git repository. To use this mode, you must add permissions for your CodeBuild environment to pull from your repository.
To add permissions to your CodeBuild service role policy, you create a customer-managed policy
that you attach to your CodeBuild service role. The following steps create a policy that specifies
the codecommit:GitPull
permission in the action
field.
To use the console to add the GitPull permissions
-
To find your CodeBuild service role, open the build project used in your pipeline and navigate to the Build details tab.
-
Choose the Service role link. This opens the IAM console where you can add a new policy that grants access to your repository.
-
In the IAM console, choose Attach policies, and then choose Create policy.
-
On the JSON tab, paste the following sample policy.
{ "Action": [ "codecommit:GitPull" ], "Resource": "*", "Effect": "Allow" },
-
Choose Review policy. Enter a name for the policy (for example,
codecommit-gitpull
), and then choose Create policy. -
Return to the page where you were attaching permissions, refresh the policy list, and select the policy you just created. Choose Attach policies.
Pipeline error: A deployment with the CodeDeployToECS action returns an error message: "Exception while trying to read the task definition artifact file from: <source artifact name>"
Problem:
The task definition file is a required artifact for the CodePipeline deploy action to Amazon ECS
through CodeDeploy (the CodeDeployToECS
action). The maximum artifact ZIP size in the
CodeDeployToECS
deploy action is 3 MB. The following error message is returned
when the file is not found or the artifact size exceeds 3 MB:
Exception while trying to read the task definition artifact file from: <source artifact name>
Possible fixes: Make sure the task definition file is included as an artifact. If the file already exists, makes sure the compressed size is less than 3 MB.
GitHub (via OAuth app) source action: Repository list shows different repositories
Problem:
After a successful authorization for a GitHub (via OAuth app) action in the CodePipeline console, you can choose from a list of your GitHub repositories. If the list does not include the repositories you expected to see, then you can troubleshoot the account used for authorization.
Possible fixes: The list of repositories provided in the CodePipeline console are based on the GitHub organization the authorized account belongs to. Verify that the account you are using to authorize with GitHub is the account associated with the GitHub organization where your repository is created.
GitHub (via GitHub App) source action: Unable to complete the connection for a repository
Problem:
Because a connection to a GitHub repository uses the AWS Connector for GitHub, you need organization owner permissions or admin permissions to the repository to create the connection.
Possible fixes: For information about permission levels
for a GitHub repository, see https://docs.github.com/en/free-pro-team@latest/github/setting-up-and-managing-organizations-and-teams/permission-levels-for-an-organization
Amazon S3 error: CodePipeline service role <ARN> is getting S3 access denied for the S3 bucket <BucketName>
Problem:
While in progress, the CodeCommit action in CodePipeline checks that the pipeline artifact bucket
exists. If the action does not have permission to check, an AccessDenied
error
occurs in Amazon S3 and the following error message displays in CodePipeline:
CodePipeline service role
"arn:aws:iam::AccountID
:role/service-role/RoleID
"
is getting S3 access denied for the S3 bucket "BucketName
"
The CloudTrail logs for the action also log the AccessDenied
error.
Possible fixes: Do the following:
-
For the policy attached to your CodePipeline service role, add
s3:ListBucket
to the list of actions in your policy. For instructions on to view your service role policy, see View the pipeline ARN and service role ARN (console). Edit the policy statement for your service role as detailed in Add permissions to the CodePipeline service role. -
For the resource-based policy attached to the Amazon S3 artifact bucket for your pipeline, also called the artifact bucket policy, add a statement to allow the
s3:ListBucket
permission to be used by your CodePipeline service role.To add your policy to the artifact bucket
-
Follow the steps in View the pipeline ARN and service role ARN (console) to choose your artifact bucket on the pipeline Settings page and then view it in the Amazon S3 console.
-
Choose Permissions.
-
Under Bucket policy, choose Edit.
-
In the Policy text field, enter a new bucket policy, or edit the existing policy as shown in the following example. The bucket policy is a JSON file, so you must enter valid JSON.
The following example shows a bucket policy statement for an artifact bucket where the example role ID for the service role is
AROAEXAMPLEID
.{ "Effect": "Allow", "Principal": "*", "Action": "s3:ListBucket", "Resource": "arn:aws:s3:::
BucketName
", "Condition": { "StringLike": { "aws:userid": "AROAEXAMPLEID
:*" } } }The following example shows the same bucket policy statement after the permission is added.
{ "Version": "2012-10-17", "Id": "SSEAndSSLPolicy", "Statement": [
{ "Effect": "Allow", "Principal": "*", "Action": "s3:ListBucket", "Resource": "arn:aws:s3:::
{ "Sid": "DenyUnEncryptedObjectUploads", "Effect": "Deny", "Principal": "*", "Action": "s3:PutObject", "Resource": "arn:aws:s3:::codepipeline-us-east-2-1234567890
", "Condition": { "StringLike": { "aws:userid": "AROAEXAMPLEID
:*" } } },codepipeline-us-east-2-1234567890
/*", "Condition": { "StringNotEquals": { "s3:x-amz-server-side-encryption": "aws:kms" } } }, { "Sid": "DenyInsecureConnections", "Effect": "Deny", "Principal": "*", "Action": "s3:*", "Resource": "arn:aws:s3:::codepipeline-us-east-2-1234567890
/*", "Condition": { "Bool": { "aws:SecureTransport": false } } } ] }For more information, see the steps in https://aws.amazon.com/blogs/security/writing-iam-policies-how-to-grant-access-to-an-amazon-s3-bucket/
. -
Choose Save.
-
After you apply the edited policy, follow the steps in Start a pipeline manually to manually rerun your pipeline.
Pipelines with an Amazon S3, Amazon ECR, or CodeCommit source no longer start automatically
Problem:
After making a change to configuration settings for an action that uses event rules (EventBridge or CloudWatch Events) for change detection, the console might not detect a change where source trigger identifiers are similar and have identical initial characters. Because the new event rule is not created by the console, the pipeline no longer starts automatically.
An example of a minor change at the end of the parameter name for CodeCommit would be changing
your CodeCommit branch name MyTestBranch-1
to MyTestBranch-2
. Because the
change is at the end of the branch name, the event rule for the source action might not update
or create a rule for the new source settings.
This applies to source actions that use CWE events for change detection as follows:
Source action | Parameters / trigger identifiers (console) |
---|---|
Amazon ECR |
Repository name Image tag |
Amazon S3 |
Bucket S3 object key |
CodeCommit |
Repository name Branch name |
Possible fixes:
Do one of the following:
-
Change the CodeCommit/S3/ECR configuration settings so that changes are made to the starting portion of the parameter value.
Example: Change your branch name
release-branch
to2nd-release-branch
. Avoid a change at the end of the name, such asrelease-branch-2
. -
Change the CodeCommit/S3/ECR configuration settings for each pipeline.
Example: Change your branch name
myRepo/myBranch
tomyDeployRepo/myDeployBranch
. Avoid a change at the end of the name, such asmyRepo/myBranch2
. -
Instead of the console, use the CLI or AWS CloudFormation to create and update your change-detection event rules. For instructions on creating event rules for an S3 source action, see Connecting to Amazon S3 source actions that use EventBridge and AWS CloudTrail. For instructions on creating event rules for an Amazon ECR action, see Amazon ECR source actions and EventBridge resources. For instructions on creating event rules for a CodeCommit action, see CodeCommit source actions and EventBridge.
After you edit your action configuration in the console, accept the updated change-detection resources created by the console.
Connections error when connecting to GitHub: "A problem occurred, make sure cookies are enabled in your browser" or "An organization owner must install the GitHub app"
Problem:
To create the connection for a GitHub source action in CodePipeline, you must be the GitHub organization owner. For repositories that are not under an organization, you must be the repository owner. When a connection is created by someone other than the organization owner, a request is created for the organization owner, and one of the following errors display:
A problem occurred, make sure cookies are enabled in your browser
OR
An organization owner must install the GitHub app
Possible fixes: For repositories in a GitHub organization, the organization owner must create the connection to the GitHub repository. For repositories that are not under an organization, you must be the repository owner.
Pipelines with execution mode changed to QUEUED or PARALLEL mode fails when run limit reached
Problem: The maximum number of concurrent executions for a pipeline in QUEUED mode is 50 executions. When this limit is reached, the pipeline fails without a status message.
Possible fixes: When editing the pipeline definition for execution mode, make the edit separately from other edit actions.
For more information about QUEUED or PARALLEL execution mode, see CodePipeline concepts.
Pipelines in PARALLEL mode have an outdated pipeline definition if edited when changing to QUEUED or SUPERSEDED mode
Problem: For pipelines in parallel mode, when editing the pipeline execution mode to QUEUED or SUPERSEDED, the pipeline definition for PARALLEL mode will not be updated. The updated pipeline definition when updating PARALLEL mode is not used in the SUPERSEDED or QUEUED mode
Possible fixes: For pipelines in parallel mode, when editing the pipeline execution mode to QUEUED or SUPERSEDED, avoid updating the pipeline definition at the same time.
For more information about QUEUED or PARALLEL execution mode, see CodePipeline concepts.
Pipelines changed from PARALLEL mode will display a previous execution mode
Problem: For pipelines in PARALLEL mode, when editing the pipeline execution mode to QUEUED or SUPERSEDED, the pipeline state will not display the updated state as PARALLEL. If the pipeline changed from PARALLEL to QUEUED or SUPERSEDED, the state for the pipeline in SUPERSEDED or QUEUED mode will be the last known state in either of those modes. If the pipeline was never run in that mode before, then the state will be empty.
Possible fixes: For pipelines in parallel mode, when editing the pipeline execution mode to QUEUED or SUPERSEDED, note that the execution mode display will not show the PARALLEL state.
For more information about QUEUED or PARALLEL execution mode, see CodePipeline concepts.
Pipelines with connections that use trigger filtering by file paths might not start at branch creation
Description: For pipelines with source actions that use connections, such as a BitBucket source action, you can set up a trigger with a Git configuration that allows you to filter by file paths to start your pipeline. In certain cases, for pipelines with triggers that are filtered on file paths, the pipeline might not start when a branch with a file path filter is first created, since this does not allow the CodeConnections connection to resolve the files that changed. When the Git configuration for the trigger is set up to filter on file paths the pipeline will not start when the branch with the filter has just been created in the source repository, For more information about filtering on file paths, see Add trigger with code push or pull request event types.
Result: For example, pipelines in CodePipeline that have a file path filter on a branch "B" will not be triggered when branch "B" is created. If there are no file path filters, the pipeline will still start.
Pipelines with connections that use trigger filtering by file paths might not start when file limit is reached
Description: For pipelines with source actions that use connections, such as a BitBucket source action, you can set up a trigger with a Git configuration that allows you to filter by file paths to start your pipeline. CodePipeline retrieves up to the first 100 files; therefore, when the Git configuration for the trigger is set up to filter on file paths, the pipeline might not start if there are over 100 files, For more information about filtering on file paths, see Add trigger with code push or pull request event types.
Result: For example, if a diff contains 150 files, CodePipeline looks at the first 100 files (in no particular order) to check against the file path filter specified. If the file that matches the file path filter is not among the 100 files retrieved by CodePipeline, the pipeline will not be invoked.
CodeCommit or S3 source revisions in PARALLEL mode might not match EventBridge event
Description: For pipeline executions in PARALLEL mode, an execution might start with the most recent change, such as the CodeCommit repository commit, that might not be the same as the change for the EventBridge event. In some cases, where a split second might be between commits or image tags that start the pipeline, when CodePipeline receives the event and starts that execution, another commit or image tag has been pushed, CodePipeline (for example, the CodeCommit action) will clone the HEAD commit at that moment.
Result: For pipelines in PARALLEL mode with a CodeCommit or S3 source, regardless of the change that triggered the pipeline execution, the source action will always clone the HEAD at the time it is started. For example, for a pipeline in PARALLEL mode, a commit is pushed, which starts the pipeline for execution 1, and the second pipeline execution uses the second commit.
Need help with a different issue?
Try these other resources:
-
Contact AWS Support
. -
Ask a question in the CodePipeline forum
. -
Request a quota increase
. For more information, see Quotas in AWS CodePipeline. Note
It can take up to two weeks to process requests for a quota increase.