none
How to copy data from non-headnode.

    Domanda

  • The cluster in my office have 50 nodes, each nodes have their own hard disk. But only the headnode connect to the internet. I want to copy the data, which is stored in other nodes not in the headnode, to my desktop. How should I do. The procedure what I do now, first copy the data to the shared folder of the headnode, then copy the data to my desktop. Is there any straight way to direct copy to data to my desktop?
    venerdì 21 agosto 2009 07:13

Risposte

  • Since you already have a shared folder on the headnode, it would be simplest if the jobs performed a final copy of the data from the compute nodes to the headnode share before they finish. Alternatively, you could create and run a separate data-consolidation job with tasks that copy the data from the compute nodes to the headnode share, and then a final task that runs on the headnode to copy the data to a share on your desktop.

    giovedì 19 novembre 2009 19:51

Tutte le risposte

  • Hi,

    you can enable the NAT on the headnode and then your nodes can access the outside of the private network.
    A more simpler way is to login to the headnode and forward you local drives, or mount a network share on the headnode of your local drive.
    Then access the node data from the headnode and copy it to the network or RDP share.

    Johannes
    JH
    lunedì 24 agosto 2009 10:10
  • Since you already have a shared folder on the headnode, it would be simplest if the jobs performed a final copy of the data from the compute nodes to the headnode share before they finish. Alternatively, you could create and run a separate data-consolidation job with tasks that copy the data from the compute nodes to the headnode share, and then a final task that runs on the headnode to copy the data to a share on your desktop.

    giovedì 19 novembre 2009 19:51