Guys, I have a problem and I've been researching a...
# ask-ai
p
Guys, I have a problem and I've been researching and I'm trying to do it myself but I haven't made any progress. My scenario is as follows, I had a VM with Airbyte installed, but it wasn't doing enough (we ran about 400Syncs per hour) and so I opted to put that on a GKE cluster. My cluster allows a maximum of 4 nodes, and even without having migrated all the loads, the utilization is already on 3 nodes. I saw that each job pod uploads approximately 4 containers, and these containers have 100m of requests each, so each pod uploads with 400mCPU of requests. The use of nodes is very low, which leads me to believe that I can reduce these requests well so that the cluster does not become oversized for nothing. I was able to edit the value of the requests field for the main container through values.yaml (I'm deploying with helm) But the other containers (relay-stderr, relay-stdout, call-heartbeat-server) continue with 100m EACH. I saw that there is a variable in the main container called: SOCAT_KUBE_CPU_REQUEST and it has a value of 0.1. I haven't been able to change the value of this variable yet. Anyway, has anyone gone through this? Thanks.
k
A message from kapa.ai
Hmm, I don't know enough to give you a confident answer yet. However, you can refer to the Airbyte documentation on [scaling Airbyte](https://docs.airbyte.com/operator-guides/scaling-airbyte) and [configuring Airbyte on Kubernetes](https://docs.airbyte.com/operator-guides/configuring-airbyte) for more information and guidance.
p
This variable SOCAT_KUBE_CPU_REQUEST don't appear on this documentation. As if she wasn't customizable