Troubleshooting Deployment Failures
Deployment could fail due to various reasons, such as:
- Incorrect CF parameters input.
- Insufficient account permissions for CF to create resources.
- Non-available resources on AWS.
- The AMI is not accessible. In this case, you will need to make sure that a defined AMI is available in the account.
- Miscellaneous errors.
Once you have encountered a failure, you will notice that the CF automatically rolls back and deletes all created resources. The automatic rollback sometimes makes it harder to debug the reason for failure.
In order to better debug the reason for failure, you may want to check the Preserve successfully provisioned resources checkbox in the CF section, as follows:
This option is useful to keep resources up even if deployment fails. It allows users to monitor stack resources and failure reasons, and access CloudWatch and Lambda logs easily via the AWS CF console web pages.
One of the typical failures for deployment is when setting an incorrect name for the S3ConfBucketName parameter. This failure is encountered only at a later deployment stage (after several minutes), and it will be shown as part of the deploy lambda resource (failure to access the S3 bucket).
To get Lightbits support for deployment failures, it is suggested to collect:
- AWS CF S3 URL (from the ‘Create Stack’ page).
- to determine the Amazon S3 URL used as the source CFT.
- Region - to determine the Region where the cluster was deployed in.
- AWS CF - screenshots or copy/paste of root stack > ’Parameters’ tabs.
- This will help with understanding the parameter inputs used.
- AWS CF - screenshots or copy/paste of root and nested stacks > ‘Events’ tabs.
- This will usually help to determine the main reason for failure.
- AWS CloudWatch - copy/paste or export the log of deploying Lambda.
- This may help to further analyze the problem, if related to Lambda.
This is the kind of initial debug information that will assist Support in analyzing your deployment issue.
Example - Incorrect S3ConfBucketName Name Failure
The following is an example of screenshot information collected for a typical failure of an incorrect S3ConfBucketName parameter input.
- Screenshot of Create Stack to obtain Amazon S3 URL used to initiate the deployment:
- The Parameters tab of the root stack (also important: the Region can be seen in the right top corner).
- The Events tab of the failing nested stack (‘lbapp’ stack in this example):
- The CloudWatch Lambda log per log link, indicated in the create_failed event (from above):
