TCP connections in the headnode RRS feed

  • Question

  • Hi,

    The Tcp connection of headnode in our environment is usually less than 10k but we occasionally saw a very high TCP connections (like 84k+) in the headnode. Then headnode went into a bad state and no response for a while. We currently have around 250+ wcfclient connect to the headnode. Do you know what maybe the cause? Is there anyway we can troubleshoot this problem?



    Wednesday, February 9, 2011 9:07 PM

All replies

  • Hi,

    Can you provide a bit more information about your cluster configuration (cluster size and network configuration, for example) and any relevant information which might have led to the "bad state"?   Is the head node also a compute node?

    Were there any jobs running when the head node became unresponsive? If so, can you provide more information about the job?

    Thursday, February 24, 2011 10:01 AM