Changes between Version 11 and Version 12 of ticket/370/ticket/370/TicketDetails/OpsReadFromObstore


Ignore:
Timestamp:
Oct 3, 2019 3:52:15 PM (2 months ago)
Author:
Jin Lee
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • ticket/370/ticket/370/TicketDetails/OpsReadFromObstore

    v11 v12  
    3535
    3636   * For obstore files which have large numbers of batches failures can occur with either Ops_CreateODB or Ops_ExtractAndProcess:
    37      * Ops_CreateODB fails and the failure is at the step where PE0 is trying to read the obstore file and distribute the data to other PE's. In this case try increasing PBS walltime request.
    38      * Ops_CreateODB fails and the failure is ????
     37     * Ops_CreateODB fails - decrease buffersize (which is roughly the number of observations in each batch) in inverse proportion to the larger number of batches
     38     * Ops_ExtractAndProcess fails - if the failure happens towards the end of the processing where updating of ODB1 takes place then increasing the number of nodes and memory can fix this problem
     39     * for some obsgroups - e.g. sonde - the number of batches used in its obstore may be unusually large. This is fixed by using nodes which have larger memories.
     40     * for obstype of satwind and surface no amount of fine-tuning allow the tasks to read all observations. It's possible the number of observations as reported by print-obstore is not correct
    3941
    4042=== Resources ===