Final EKS Auto Mode testing #178
Conversation
| # # security group that Karpenter should utilize with the following tag | ||
| # # (i.e. - at most, only one security group should have this tag in your account) | ||
| # "karpenter.sh/discovery" = local.name | ||
| "eks.amazonaws.com/discovery" = "modern-engineering" |
There was a problem hiding this comment.
keep local.name instead of modern-engineering.
…le since Hari's PR will take care of that
There was a problem hiding this comment.
LGTM. Tested with new 1.32 version locally. works fine. tested multiple times. Time to test in TE once merged @shapirov103 @elamaran11
|
QQ, We are using Auto Mode only on Management cluster or also on Child clusters? |
|
I will do a code review tomorrow. |
Only Management cluster. Child clusters are using blueprints still |
There was a problem hiding this comment.
@Eli1123 @hmuthusamy Firstly great work, Im good with the changes, i dont see any issues on the changes. What im concerned on is the below which we need to prove that there is no issues before we can merge:
- Auto Mode runs only on BottleRocket, did we check if each any every workload, argocd app that was deployed on Management cluster works fine and shows Green in Argo UI?.
- Did all the pods come up fine in management cluster or we see any issues with any pods. Are there any pods that cannot work on BR, we need to double check this
- Did any one do apple to apple comparison of TE in Workshop vs the deployment run with Auto Mode in your environment. Do you see any difference in behavior
- Did we have anyone run the workshop like run end to end with your auto mode environment. This is mandatory for merge.
This is a breaking change, so we need to have thorough review, before we can merge.
@shapirov103 Please check from your end too.
|
@elamaran11 Just chatted with Hari and a brief review found some possible issues to work through. We plan to meet up tomorrow to trouble shoot more as his proper wifi should return later today, and he will run end to end one more time as well to see if he replicates my possible issues. |
|
Hari identified the initial issue to be related to load balancer security group issues with EKS Auto mode. Currently I've been trouble shooting how to resolve that and then see if any other errors surface with auto mode. |
Currently was able to have it deploy to my environment and work properly utilizing EKS auto for the management cluster and resolved the EBS volumes issues with changes to the postgres.yaml file and adding the keycloak.yaml file.
Passing to Hari to test changes locally to confirm.