AWS Codedeploy Wordpress deployment fails due to PHP version change in buildspec file - wordpress

Question

I am running my Wordpress site hosted in AWS EC2 [Amazon Linux2 AMI].
I have a pipeline to deploy the latest source from Github.
The codebuild buildspec file is as follow: [Mentioned only related part]
version: 0.2
phases: install:
runtime-versions: php: 7.4
Codebuild project in cloudformation is like this:
Image: 'aws/codebuild/amazonlinux2-x86_64-standard:3.0'
Using above configuration, deployment is going without problem and problem happens when I update like below: [Mentioned only related part]
In cfn template:
Image: 'aws/codebuild/amazonlinux2-x86_64-standard:4.0'
In buildspec: runtime-versions: php: 8.1
Validation script in deployment hooks runs some validation commands to check RDS connection and it fails:
LifecycleEvent - ValidateService Script - deploy_hooks/validate_service.sh [stdout]OK SO FAR
[stderr]mysqlcheck: Got error: 1045: Access denied for user 'root'#'10.37.253.207' (using password: NO) when trying to connect
After deployment, I loginto the ec2 instance and below WP cli command fails as well:
$ wp db check
These above problems does not occur when I roll back to previous state of the PHP version.
In cfn template:
Image: 'aws/codebuild/amazonlinux2-x86_64-standard:3.0'
In buildspec:
runtime-versions: php: 7.4
What might be the problem ?
Reference:
https://docs.aws.amazon.com/codebuild/latest/userguide/build-env-ref-available.html
https://docs.aws.amazon.com/codebuild/latest/userguide/available-runtimes.html#linux-runtimes
** UPDATE:
Codedeploy is failing to deploy DB endpoint , password and other information in wp-config.php file when codedeploy is executed by codepipeline.
If I terminate the ec2 instance, the auto-scaling will launch a new instance for me and deployment is triggered by the codedeploy and the deployment is success. I can access the site.
Does it mean that executing Codedeploy through Codepipeline is the problem?