Re: [ns] memory leak problem

[Eric Weigle <ehw@lanl.gov>]

Hello-

I was never able to track down all the problems with memory that I was having.
One of my simulations (coincidentally, with 1000 one-way TCP flows) always ended
up using all my memory and 2 gigs of swap about 3/4 of the way through a 200s
simulation, then dying.

I managed to get results by running the exact same simulation with ns in gdb. If
you don't care about speed or convenience (shell scripts/etc) you might give
that a try. For whatever reason (gdb doing garbage collection? Obscure timing
issues? I don't know...) simulations which died running natively ran fine and
gave results when run in gdb. Worth a try, if you haven't.

If I ever get my hands on a solaris box around here, I'd just use a tool like
purify to track any leaks... unfortunately I don't know of a good tool for
Linux/x86 that does the same thing and works with C++. (Assuming the error is in
C++...) If anybody does know of such a beast, let me know.


-Eric

--------------------------------------------
 Eric H. Weigle   CCS-1, RADIANT team
 ehw@lanl.gov     Los Alamos National Lab
 (505) 665-4937   http://home.lanl.gov/ehw/
--------------------------------------------