Issue with resource requests not updating in Airbyte pod after upgrade to V1.0.0 on EKS with helm charts

Slack Knowledge Base Q&A Platform Questions
, , , ,
1

Summary

After upgrading to V1.0.0 on EKS with helm charts, noticed that replication jobs are using more capacity due to 3 containers in the same pod. Resource requests are not updating in the pod despite changing values for JOB_MAIN_CONTAINER_MEMORY_REQUEST and JOB_MAIN_CONTAINER_CPU_REQUEST.


Question

Hi all,
I’m using an OSS instance deployed on EKS with helm charts.
after the upgrade to V1.0.0 i’ve noticed that the replication jobs are using more capacity (memory and cpu) because there are now 3 containers (replication/source/destination) in the same pod, but when trying to change the resource requests, while the 2 env vars are updated correctly: JOB_MAIN_CONTAINER_MEMORY_REQUEST and JOB_MAIN_CONTAINER_CPU_REQUEST but the actual requests in the pod itself aren’t changing
those are my values:


      edition: community

      jobs:
        resources:
          limits:
            cpu: 1
            memory: 3Gi
          requests:
            cpu: 500m
            memory: 2Gi```

<br>

---

This topic has been created from a Slack thread to give it more visibility.
It will be on Read-Only mode here. [Click here](https://airbytehq.slack.com/archives/C021JANJ6TY/p1727353179181999) if you want 
to access the original thread.

[Join the conversation on Slack](https://slack.airbyte.com)

<sub>
["eks", "helm-charts", "airbyte-pod", "resource-requests", "replication-jobs"]
</sub>
2

Yes, thanks.

You said the JOB_MAIN_CONTAINER_MEMORY_REQUEST env var was updated correctly. Where did you see that? In the config map, or on the deployments?

3

first of all, thank you!

both actually. you can see in the file i attached
and i also see it in the configmap

4

In the past, we’ve noticed that changing these resource values on an existing cluster doesn’t alway cause the various components (server, workload-launcher, worker) to restart, so they don’t pick up the changes. So you could try deleting the server, workload-launcher, and worker pods (they’ll get recreated automatically) to see if that helps.

5

yeah, that makes sense. i noticed already that the workload-launcher didn’t pickup the image changes and i delete the deployment and it worked after it was redeployed
i’ll try that now

thanks!

6

i deleted the server deployment and reconcile the flux hr and it works now
thank you! :slightly_smiling_face:

7

Can you show the output of kubectl get pod for one of the job pods?