Dump a Non-Compute Node

Trigger a non-compute node (NCN) memory dump and send the dump for analysis. This procedure is helpful for debugging NCN crashes.

Prerequisites

An NCN has crashed or an administrator has triggered a node crash.

Procedure

  1. (ncn#) Force a dump on an NCN.

    echo c > /proc/sysrq-trigger
    
  2. Wait for the node to reboot.

    The NCN dump is stored in /var/crash on the local disk after the node is rebooted.

  3. (ncn#) Collect the dump data using the System Diagnostic Utility (SDU).

    Refer to the “Run a Triage Collection with SDU” procedure in the SDU product stream documentation for more information about collecting dump data.

    The --start_time command option can be customized. For example, -1 day, -2 hours, or a date/time string can be used. For more information on the SDU command options, use the sdu --help command.

    sdu --scenario triage --start_time DATE_OR_TIME_STRING --end_time DATE_OR_TIME_STRING --plugin ncn.gather.nodes --plugin ncn.gather.kernel.dumps
    

    Refer to the SUSE documentation for more information on memory dumps or crash dumps.