memory leak with HP tools and Linux

[Chuck5]

I have a memory leak on G5s running the HP management tools. These are all running CentOS 4.4 (or the identical Redhat ES4 Update 4 in a few cases), with kernel 2.6.9-42. Has anyone else experienced the same thing?

If the HP tools are installed (hpasm and hprsm packages), and especially if the machine has a lot of free memory, memory vanishes at a slow but steady rate.

I am currently testing 2.6.9-55.0.2, with the latest hpasm and hprsm packages. It takes a few days for the leak to show itself. But neither Redhat nor HP lead me to believe that the latest packages fix anything like this.

This has become a huge problem for me, as I have to choose between rebooting machines weekly, or not having monitoring of hardware failures.

I do not have the problem with a later 2.6 kernel, e.g. the one that comes with CentOS 5.
Of course the problem goes away if I uninstall the HP tools.

Here is an example:

# uname -a
Linux host1 2.6.9-42.ELsmp #1 SMP Tue Aug 15 10:35:26 BST 2006 x86_64 x86_64 x86_64 GNU/Linux

# rpm -q hpasm hprsm
hpasm-7.7.0-115.rhel4.x86_64
hprsm-7.7.0-114.rhel4.x86_64

# free
total used free shared buffers cached
Mem: 32911228 14501672 18409556 0 128732 782808
-/+ buffers/cache: 13590132 19321096
Swap: 4194296 0 4194296

Nowhere near 13GB is actually in use:
# ps auxw|awk '{A+=$6}END{print A}'
257416

(no shared memory segments or tmpfs usage on this machine either).

[David Claypool]

This needs to be reported officially to HP Customer Services. In the US, 1-800-HP=INVENT and say ProLiant during the prompts.

[Brian Paxton]

I have a similar problem. A machine running RHEL4 with hpasm 7.7.0, and it is memory leaking about 1 MByte a day.

However, I have another machine as follows:

Proliant DL380G4
Red Hat Enterprise Linux Server release 5.1 (Tikanga)
Linux Kernel 2.6.18-53.el5 #1 SMP Wed Oct 10 16:34:19 EDT 2007 x86_64 x86_64 x86_64 GNU/Linux

hp-ilo-8.0.0-144.rhel5
hpasm-8.0.0-173.rhel5
hponcfg-1.7.0-2
hpacucli-8.0-14

On this machine, cmanicd is leaking 2 Mbytes of RAM a day:

1 day after reboot: cmanicd VSZ=133588k RSS=5408k
2 days after reboot: cmanicd VSZ=135600k RSS=7396k
3 days after reboot: cmanicd VSZ=137364k RSS=9248k
6 days after reboot: cmanicd VSZ=143644k RSS=15424k
7 days after reboot: cmanicd VSZ=145272k RSS=17056k

I have witnessed this application consuming 300 MBytes of RAM.

I think there is a serious issue with this software. I recommend running version 7.3, as it is the only version I have found which appears to be stable.