I have 4GB RAM VM with trisul (and ntopng, but that's different story)
After some time of leaving web admin open in browser, i see insane usage of RAM by webtrisul threads, with single thread taking 70% of RAM (!). All other threads dying except that one fat bastard - it continues to grow and i see several oom-killer notifies in dmesg.
Typical scenario:
[159420.803418] ruby invoked oom-killer: gfp_mask=0x280da, order=0, oom_adj=0, oom_score_adj=0
[159420.803421] ruby cpuset=/ mems_allowed=0
[159420.803423] Pid: 7100, comm: ruby Tainted: G O 3.2.0-54-generic #82-Ubuntu
[159420.803425] Call Trace:
lex@dpi >>> free -m
total used free shared buffers cached
Mem: 3954 3759 195 0 4 42
-/+ buffers/cache: 3713 241
Swap: 2043 2033 10
Sometimes it becames so slow with full SWAP and RAM so the only real way is to physically (well, sorta) reboot that VM.
BTW, trisuld process still works good with around 10% RAM usage.