Hello everyone,

I have a Fujitsu-siemens server (4 cores, 6GiB RAM) with RHEL 4 (2.6.9-78.0.13.ELsmp #1 SMP) and I am very confused about last month's memory consumption. I monitor the server with SNMP, and the memory performance graphs shows a slow increment of used memory, as if it were a mem leak, and a slow decrement of cached memory. The standards for the server (running two web servers and one postgresql) are about a 10% of used memory, and the rest in cache, and now used memory is about 80% of the total (+4 GB), and almost all the rest cached.

The weird thing is that when I try to locate the process eating all the memory, I find nothing. Some data follows:

ps_mem.py out (three last lines, the more relevant):
-------------
2.5 MiB + 1.0 MiB = 3.6 MiB nginx (3)
2.2 MiB + 2.4 MiB = 4.5 MiB snmpd
97.1 MiB + 6.6 MiB = 103.7 MiB httpd (119)
----------
free -m out:
-------------------
total used free shared buffers cached
Mem: 6084 6021 63 0 51 1512
-/+ buffers/cache: 4457 1627
-------------
ps, top, htop are showing numbers that added are like the ones from ps_mem.py (aprox.)
Calculating the totals using smaps.pl and pmap against all alive pids (scripting with awk and ruby) is not helping, they show similar numbers (not even near 1GiB of used memory, taking account of shared mem)

I also checked for rootkits (chkrootkit, rkhunter) just in case, and I don't know what more I can do, can anyone help me? Am i forgetting something?

P.S. Sorry for my bad english.