• Main INDEX
  • Monthly INDEX
  • PREV
  • NEXT
    Make New Entry, Make Followup Entry

    User name R. Michaels

    Log entry time 12:27:35 on May 17, 2009

    Entry number 274160

    This entry is a followup to: 274123

    Followups:

    keyword=retune of parameter in pingpong script

    I tuned the parameter for the "available size of disk" in the pingpong script on a-onl
    to avoid the situation of running out of disk space. When a run is started, or when a
    file-splitting occurs, this script (pingpong) is supposed to decide what is the next
    available disk with enough space for a data file (4 GB). Well, it seems the problem was
    that only 134 GB is available on a disk, not 141 GB. Changed 141 to 134 in the script.

    Also, it seems disk4 has gotten into a state where it's nearly always full. I don't see anything
    wrong with this, but will see if I can clean it up when I'm on day shift tomorrow. DO NOT
    try to clean disks by hand. It is very dangerous. All this is done automatically and carefully
    by scripts.

    I think what's happened is this: the script that deletes files has it's "aggressiveness" tuned
    to the amount of disk space available globally; if lots of disk is available the data hangs
    around for a week, and if the disks are very full the data hangs around for 8 hours (of course
    they must be in MSS in duplicate with correct byte count). And so ... if one disk become full
    for some rare reason, it does not get cleaned aggressively if the others are not also full.
    This situation, together with the "bug" fixed in paragraph 1, caused all the trouble, it seems.

    The suggestion to "write to MSS more often" is not relevant. We already are.



    A copy of this log entry has been emailed to: moffit