I'm not sure if the following is really a memory leak or some kind of bug, so I would like to know the opinion of the developers first before filing a bugreport.
1) my setup
I'm using the latest 64bit 0.9dev LuxRender build for Linux: gcc453-x86_64-sse2-120406-lux-v09dev-120406-14d3aa545f4c
running on 2 Debian Linux nodes: totoro-slv-002 acts as server (luxconsole -s) and on totoro-slv-001 I start luxconsole like this: "nice luxconsole -i 240 -r -u totoro-slv-002 luxtest.lxs"
2) the issue
On both nodes luxconsole increasingly uses up memory until they start swapping.
However, when I kill the running luxconsole processes and restart them resuming with the same .flm file then the memory usage drops to normal again.
I first thought that the increasing memory usage may be related to the amount of calculated samples, but then I wonder why resuming a flm consumes the same amount of memory than a fresh start from 0s/p?
I'll give temporarily access to my server's munin monitoring graphs which should visualize the issue:
Especially the memory/swap graphs show low usage when starting luxconsole and a increase over time until swapping. Then - when I kill the processes and resume the flm again - you see a drop in memory usage/swapping.
Thanks for explanation what this could be and if there's a workaround or fix.