Uploaded image for project: 'Migration Toolkit for Virtualization'
  1. Migration Toolkit for Virtualization
  2. MTV-4482

MTV 2.10.x forklift-cli-download pod is `OOMKilled` due to insufficient default memory limits operator prevents persistent override (regression from 2.9)

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Major Major
    • None
    • 2.10.0
    • Operator
    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • True

      Description of problem:

      • After upgrading the Migration Toolkit for Virtualization (MTV) Operator from version 2.9.x to 2.10.2 / 2.10.3, the forklift-cli-download deployment consistently enters a CrashLoopBackOff state.
      • The container is being terminated with exitCode: 137 (OOMKilled) due to insufficient default memory limits (64Mi). As a result, the deployment never reaches the expected replica count, and MTV reports that cluster infrastructure is unhealthy.
      • While increasing the memory limits on the forklift-cli-download deployment resolves the issue, the deployment is fully managed by the MTV Operator and any manual changes are reverted. Additionally, changes made to the ForkliftController custom resource are also reverted by the operator, leaving no supported mechanism to persist resource overrides.
      • This behavior represents a regression from MTV 2.9.x, where the same environment and workloads function correctly without OOM events.
      • Observed one important thing is that in MTV 2.9.x version there is no forklift-cli-download pod is present. 

      Version-Release number of selected component (if applicable):

      - OpenShift Version: 4.20.x
      
      - MTV Operator Version: 2.10.2 or 2.10.3

      How reproducible:

      Always / N%
      

      Steps to Reproduce:

      1.
      2.
      3.
      

      Actual results:

      • `forklift-cli-download` pod enters `CrashLoopBackOff`.
      • Pod is repeatedly terminated with
      Reason: OOMKilled 
      Exit Code: 137  
      • Deployment never reaches expected replicas
      • Manual patches to Deployment or ForkliftController are reverted by the operator

      Expected results:

      • forklift-cli-download pod should start successfully with default resource settings OR
      • Operator should provide a supported mechanism (via ForkliftController or configuration) to override resource requests/limits
      • MTV infrastructure should reach a healthy state after operator installation or upgrade.

      Additional info:
      Example pod status:

        containerStatuses:
        - allocatedResources:
            cpu: 50m
            memory: 64Mi
          containerID: cri-o://25a6be63756814621a8bf76c4e33d27f0cac95736f997b1f6f5afbb170a71bdf
          image: registry.redhat.io/migration-toolkit-virtualization/mtv-cli-download-rhel9@sha256:768422ea04397f8e9e5b588d14671568ae0da0bd6a39668abd77948447bb9fcf
          imageID: registry.redhat.io/migration-toolkit-virtualization/mtv-cli-download-rhel9@sha256:768422ea04397f8e9e5b588d14671568ae0da0bd6a39668abd77948447bb9fcf
          lastState:
            terminated:
              containerID: cri-o://25a6be63756814621a8bf76c4e33d27f0cac95736f997b1f6f5afbb170a71bdf
              exitCode: 137
              finishedAt: "2026-02-02T07:59:41Z"
              reason: OOMKilled
              startedAt: "2026-02-02T07:59:33Z"
          name: forklift-cli-download
          ready: false
          resources:
            limits:
              cpu: 100m
              memory: 128Mi
            requests:
              cpu: 50m
              memory: 64Mi
          restartCount: 31 

              mnecas@redhat.com Martin Necas
              rhn-support-hthakare Harshal Thakare
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated: