Uploaded image for project: 'Marathon'
  1. Marathon
  2. MARATHON-8118

CPU resource specification for second (and onward) container ignored usage when adding a pod.

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Blocker
    • Resolution: Won't Do
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: API
    • Labels:

      Description

      Overview

      Currently, the the value for the field `.containers[1].cpus` is ignored when creating a new pod, the default of "one" being used, instead.

      To reproduce, post ucr-registry-cert.json to the pods API:

      cat ucr-registry-cert.json | http :8080/v2/pods
      

      Then, look at the CPUs value for the second container. Compare with the value from the original.

      I have confirmed that the fields "mem", and "disk" are in fact respected, but "cpus" is indeed ignored (for the second container resources specification) click save click save.

      Original issue

      When using dcos marathon pod add ucr-registry-cert.json, where the push-image container uses 0.01 CPU, the resulting pod definition on DC/OS shows 1 CPU:

        Attachments

          Activity

            People

            • Assignee:
              icharalampidis Ioannis Charalampidis
              Reporter:
              chun-hunghsiao Chun-Hung Hsiao
              Team:
              Orchestration Team
              Watchers:
              Chun-Hung Hsiao, Gastón Kleiman, Ioannis Charalampidis, Tim Harper
            • Watchers:
              4 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: