Uploaded image for project: 'Mesos'
  1. Mesos
  2. MESOS-9017

Persistent storage and reserved resources

    XMLWordPrintableJSON

Details

    • Task
    • Status: Resolved
    • Minor
    • Resolution: Invalid
    • 1.5.0
    • None
    • allocation, framework, storage
    • None

    Description

      I'm trying to make a private dedicated slave to one docker container of elasticsearch. This slave also has a dedicated disk that I want to make it available to the container with "mount" type persistent volume.

      I use static reserved resources which the agent is able to load and are available.

      cat /etc/default/mesos-slave
      MASTER=`cat /etc/mesos/zk`
      MESOS_RESOURCES='[{"name":"ports","ranges":{"range":[

      {"begin":30000,"end":32000}

      ]},"type":"RANGES","role":"*"},{"name":"cpus","type":"SCALAR","scalar":

      {"value":4}

      ,"role":"*"},{"name":"cpus","reservations":[

      {"role":"eh","type":"STATIC"}

      ],"scalar":{"value":4},"type":"SCALAR"},{"name":"mem","reservations":[

      {"role":"eh","type":"STATIC"}

      ],"scalar":{"value":15023},"type":"SCALAR"},{"name":"mem","type":"SCALAR","scalar":

      {"value":15023}

      ,"role":"*"},{"disk":{"source":{"mount":

      {"root":"/data"}

      ,"type":"MOUNT"}},"name":"disk","reservations":[\{"role":"eh","type":"STATIC"}],"scalar":{"value":99000},"type":"SCALAR"},{"name":"ports","ranges":{"range":[

      {"begin":9100,"end":9400}

      ]},"reservations":[\{"role":"eh","type":"STATIC"}],"type":"RANGES"}]'
      MESOS_ROLES="eh"
      MESOS_DEFAULT_ROLE="eh"

       

      When trying to create the app with "acceptedResourceRoles": ["*","eh"]: "A resident app must have `acceptedResourceRoles = ["*"]` ".

      Why is the slave not private ? I'm not using any "slave_public" parameter set.

      If I use "acceptedResourceRoles": ["*"]: the offer is declined (even with logging at level3) it doesn't say why it's doing that: "Processing DECLINE call for offers: [ 4c1cdf04-d664-4c23-9762-da8df8031716-O280 ] for framework d1047c54-d05d-44fb-92d4-be7b462565c7-0000 (marathon) at scheduler-186208c1-08c7-452e-922a-cfbdeedd4ca4@10.101.0.211:34634"

       

      Is there any way on how to manage this kind of deployment ?

       

      Attachments

        Activity

          People

            Unassigned Unassigned
            adrian.paraschiv Adrian Paraschiv
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: