Hello OSSEC users,
I really appreciate your input on this ...
 
My new OSSEC 2.1 installation with 10 clients, is seeing ...
  1) High CPU usage from ossec-analysisd (may consume an entire CPU).
  2) Agents reporting "Disconnected" en-mass, every couple of days.
 
And wondering ...
  1) Is it normal for ossec-analysisd to peg an entire CPU ?
  2) Does re-occurance of agent "Disconnected" indicate that my server is near capacity for clients ?
  3) What is a typical server architecture/model for 10, 100 and 1000 clients ?
 
It seems that 10 clients is a very low number to be experiencing performance limitations, and wonder how many clients people are commonly supporting on a single OSSEC server.
 
Also surmizing that since ossec-analysisd and ossec-syscheckd (the heavy hitters) are single threaded, the new multicore/multithreaded CPU's will offer little benefit.
 
 
Server config ...
  Sun V210
  2 SPARC processors at 1002 Mhz.
  8GB memory
  Solaris 10
 
Client config ...
  Solaris 8 and 10.
  Previous generation SPARC
 
OSSEC version 2.1 .
Nearly all default settings.
Active alert disabled.
 
 
Normal server during a client syscheck scan (forwarding database phase) ...
 
# prstat ...
   PID USERNAME  SIZE   RSS STATE  PRI NICE      TIME  CPU PROCESS/NLWP      
 25049 ossec    4496K 3224K sleep    0   19   0:18:24  29% ossec-analysisd/1
 25060 ossecr   3720K 2288K sleep    2   19   0:00:21 0.2% ossec-remoted/3
 25067 ossec    3048K 1568K sleep    2   19   0:00:00 0.0% ossec-monitord/1
 25053 root     2864K 1592K sleep    2   19   0:00:00 0.0% ossec-logcollec/1
 25038 ossecm   3056K 1656K sleep    2   19   0:00:00 0.0% ossec-csyslogd/1
 25041 ossecm   4560K 2736K sleep    2   19   0:00:00 0.0% ossec-maild/1
 25063 root     3064K 1560K sleep    2   19   0:00:00 0.0% ossec-syscheckd/1
 
 
Overloaded server during a client syscheck scan (possibly several simultainious scans) ...
 
# ./agent_control -l 
OSSEC HIDS agent_control. List of available agents:
   ID: 000, Name: ux442 (server), IP: 127.0.0.1, Active/Local
   ID: 001, Name: uxlab5, IP: 10.32.2.125, Disconnected
   ID: 002, Name: ux142, IP: 168.230.131.234, Active
   ID: 003, Name: ux145, IP: 168.230.140.157, Disconnected
   ID: 004, Name: ux123, IP: 168.230.129.94, Disconnected
   ID: 007, Name: ux141, IP: 168.230.129.174, Disconnected
   ID: 010, Name: ux144, IP: 168.230.129.179, Disconnected
   ID: 011, Name: ux143, IP: 168.230.131.236, Active
   ID: 012, Name: ux138, IP: 168.230.131.205, Disconnected
   ID: 013, Name: ux139, IP: 168.230.131.82, Active
   ID: 014, Name: ux140, IP: 168.230.129.205, Disconnected
List of agentless devices:

# prstat ...
   PID USERNAME  SIZE   RSS STATE  PRI NICE      TIME  CPU PROCESS/NLWP      
 13885 ossec    4648K 3344K cpu0     0    0    2:39:24  49% ossec-analysisd/1
 13897 ossecr   3736K 2368K sleep   59    0   0:08:42 0.0% ossec-remoted/3
 ...
Total: 74 processes, 168 lwps, load averages: 1.14, 1.04, 0.99
 
 
All comments and experiences welcome.
Thank You,
Ken Wachtler
New OSSEC user, Minnesota USA

Reply via email to