Changes between Version 1 and Version 2 of Ticket #5102, comment 11


Ignore:
Timestamp:
Jan 17, 2013, 10:02:24 PM (9 years ago)
Author:
x190
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #5102, comment 11

    v1 v2  
    88In general, I think F_NOCACHE on Darwin is only applicable to when you open a file. Turn it ''off'' if you need to re-use the data or ''on'' if data is only needed once. Therefore, I would suggest removing it completely from tr_close_file() where, if it has any effect at all, it is giving the wrong "hint" to the OS.
    99
    10 Okay, download is now complete. About a minute after the transfer switched to seed mode, memory usage increased and disk reads are roughly equivalent to upload speed. VM size dropped (by 3 GB), Free Memory ''dropped'' and continues to drop, and Inactive Memory ''increased'' and continues to increase, all pretty much the opposite of what you would want to achieve, I guess.
     10Okay, download is now complete. About a minute after the transfer switched to seed mode, memory usage increased and disk reads are roughly equivalent to upload speed. VM size dropped (by 3 GB), Free Memory ''dropped'' and continues to drop ''considerably'', and Inactive Memory ''increased'' and continues to increase, all pretty much the opposite of what you would want to achieve, I guess.
     11
     12I'm wondering if F_NOCACHE is used appropriately vis-à-vis opening a file for seeding? Is its use in tr_set_file_for_single_pass() always appropriate? There does appear to be memory issues with seed mode that are not seen during downloading. I don't think complaints about 'inactive' memory are valid anymore even if they were semi-valid years ago and should not, in themselves, have anything to do with application caching management.
     13
     14