Resource ratio updates in UMA deployment

Products

• DX SaaS

Issue/Introduction

We have several pods in our Dev OpenShift cluster that keep crashing due to resource limits / requests.
Example:

ReplicaSet
Generated from replicaset-controller
Error creating: pods is forbidden: [cpu max limit to request ratio per Container is 4, but provided ratio is 10.000000, memory max limit to request ratio per Container is 2, but provided ratio is 10.000000]

This appears to have been caused the the UMA autoattach (logs attached).

Where in the deployment can we make the adjustments, and would using the below values cause an issue with the UMA?

Compliant values:
 
  • CPU: request  25m , limit  100m  → ratio = 4x (at max allowed)
  • Memory: request  50Mi , limit  100Mi  → ratio = 2x (at max allowed)

Resolution

 modify the caaiops-probe-autoattach-configmap for this sidecar contianer.

1. java-sidecarconfig.yaml

2. mule-sidecarconfig.yaml

3. nginx-sidecarconfig.yaml

4.nodejs-sidecarconfig.yaml

5. php-sidecarconfig.yaml

Ex: nodejs-sidecarconfig.yaml. Highlighted the portion that needs to be changed.

nodejs-sidecarconfig.yaml: |
containers:

• name: nodejs-probe-autoattach
  image: example.example.com/ca-uma-agent:26.1.1.35
  imagePullPolicy: IfNotPresent
  command: ["sh", "-c", "/nodeprobe/probe-init-script.sh"]
  resources:
  requests:
  memory: 50Mi
  cpu: 25m
  limits: {Modified this section to satisfy the LimitRange constaring of 4:2 for cpu and memory}
  memory: 100Mi
  cpu: 100m
  volumeMounts:

  • name: probe-shared-files
    mountPath: /nodejs-probe

• name: application-container
  env:

  • name: COLLECTOR_AGENT_HOST
    valueFrom:
    fieldRef:
    fieldPath: spec.nodeName

  • name: COLLECTOR_AGENT_PORT
    value: "5005"

  • name: CA_APM_PROBENAME
    valueFrom:
    fieldRef:
    fieldPath: metadata.name

  • name: CA_APM_APPNAME
    valueFrom:
    fieldRef:
    fieldPath: metadata.name

  • name: LOG_ENABLE_CONSOLE_MODE
    value: "enabled"

  • name: apmenv_introscope_agent_pod_namespace
    valueFrom:
    fieldRef:
    fieldPath: metadata.namespace

  • name: apmenv_introscope_agent_pod_name
    valueFrom:
    fieldRef:
    fieldPath: metadata.name

  • name: apmenv_introscope_agent_pod_ipaddress
    valueFrom:
    fieldRef:
    fieldPath: status.podIP

  • name: NODE_OPTIONS
    value: "-r /nodejs-probe/nodeprobe/node_modules/ca-apm-probe/bin/nodejs-instrumentation.js"

  resources:
  requests:
  memory: 50Mi
  cpu: 25m
  limits: {Modified this section to satisfy the LimitRange constaring of 4:2 for cpu and memory}
  memory: 100Mi
  cpu: 100m
  volumeMounts:

  • name: probe-shared-files
    mountPath: /nodejs-probe

volumes:

• name: probe-shared-files
  emptyDir: {}