PDA

View Full Version : CPD is eating memory



rubella
2010-03-05, 06:37
Hi folks

I have an annoying issue: The CPD process on one of my UTM-1 clusters (2 gateways) is eating memory. Approx. every 5 minutes, it consumes another couple of kB's. An this goes on... and on!

Finally, the real memory consumption of the CPD-process is more then 800MB after 3 months. After a reboot, cpstop/cpstart, ... , everything starts again.

Version running on this UTM box: R65 HFA50.

When I check the cpd.elg file, the only thing I see:
---> Entering addon cts_handler [LSMServerAddon, V_1.0]
<--- Exiting addon cts_handler [LSMServerAddon, V_1.0]
---> Entering addon cts_handler [LSMServerAddon, V_1.0]
<--- Exiting addon cts_handler [LSMServerAddon, V_1.0]
---> Entering addon cts_handler [LSMServerAddon, V_1.0]
<--- Exiting addon cts_handler [LSMServerAddon, V_1.0]
---> Entering addon cts_handler [LSMServerAddon, V_1.0]
<--- Exiting addon cts_handler [LSMServerAddon, V_1.0]

Every time the memory is increasing, another entry is added to the log-file...

When checking the cpd_sched_conf, the only 2 processes it's initiating are ntp and logrotate...

Does someone has an idea what is going on here?

rubella
2010-03-05, 06:39
I forgot to mention: The CPD memory usage is increasing on both members of the same cluster... Not only on the active or standby.

serlud
2010-03-06, 03:38
I forgot to mention: The CPD memory usage is increasing on both members of the same cluster... Not only on the active or standby.

Memory leak is known problem , will be never resolved (but will be every time improved) >> you should install an every HFA (install latest HFA is CP recommendation) and you will not see this memory usage icreasing.
Now CP produce an R65 HFA70 EA, to be sure you make reboot of all our systems.

In the future we will recieve an 120 HFA in every year, just to be sure all systems are will be rebooted every 3 days..

rubella
2010-03-10, 10:46
Memory leak is known problem , will be never resolved (but will be every time improved) >> you should install an every HFA (install latest HFA is CP recommendation) and you will not see this memory usage icreasing.
Now CP produce an R65 HFA70 EA, to be sure you make reboot of all our systems.

In the future we will recieve an 120 HFA in every year, just to be sure all systems are will be rebooted every 3 days..

Hi Serlud

I like your post. 120 HFA's each year, a reboot every 3 days. That would be job-protection! :-)

Anyhow, Checkpoint confirmed yesterday they found a memleaking problem with CPD (LSMAgentAddon component)

This is the procedure how to disable/delete the component (if you do not use LSM ofcourse ;))

a) check the list of CPD’s Addons
# cpd_admin list

you should see “LSMAgentAddon” in the list

b) get (and save) the list of Addons and their details
# cpd_config -p

c) delete the suspected “LSMAgentAddon” Addon
# cpd_config -d LSMAgentAddon


Be carefull, this not a fix, only a warkaround to avoid the memleaking. I hope the fix will be included in one of the next HFA's.

Have fun!