Regarding that leak - an FYI

[Rick Irving]

Modern code, 2.3.1 stable, 9-08, cleaned up most warnings.....

5 hours test time:

innfeed remains about the same at 14M of memory,
however, 5 hours later,  innd has grown to 363M,
from an initial 350M.

  It is -looking- like innd. However, the test needs to continue,
at this point.  This growth could be a natural, and relate to
load....

   am not sure though... I see swap space growth....

Hrmm....

Swap partition B (default swap):
    In-use space:           32878 pages ( 66%)
    Free space:             16247 pages ( 33%)

Swap partition A:
    In-use space:           33193 pages ( 55%)
    Free space:             27131 pages ( 44%)

Shutting down innfeed released nothing.....

Hard sig 9 to innd.... boom.

  All the swap space is back.


Swap partition B (default swap):
    In-use space:            1922 pages (  3%)
    Free space:             47203 pages ( 96%)

Swap partition A
    In-use space:            2057 pages (  3%)
    Free space:             58267 pages ( 96%)


Total swap allocation:
    Reserved space:         48007 pages ( 43%)
    In-use space:            3979 pages (  3%)
    Available space:        61442 pages ( 56%)

 Well, thats it kido's, its innd.

restarting it, my swap space remains the same.

I do -not- like the possibility that this could
relate to the realloc.....

  :(

I can set some kernel flags to "more aggressively" release
the realloc space, but, I suspect it isn't going to help.