Error in cmnetd process

[Arturo Perez del Galleg]

Hi all!
Periodically i have an error in syslog file about the cmnetd process:
Dec 30 07:25:09 atria cmnetd[4483]: DLPI error 4, unix error 233 sending to 0019bbeb9f29aa080009167e
Jan 3 08:11:51 atria cmnetd[4483]: DLPI error 4, unix error 233 sending to 001a4bf53a24aa080009167e
Mar 20 07:37:32 atria cmnetd[2722]: DLPI error 4, unix error 233 sending to 001a4bf53a24aa080009167e
Sep 7 07:36:04 atria cmnetd[2706]: DLPI error 4, unix error 233 sending to 001a4bf53a24aa080009167e
Sep 7 08:14:36 atria cmnetd[2706]: DLPI error 4, unix error 233 sending to 001a4bf53a24aa080009167e
Oct 2 08:13:33 atria cmnetd[2706]: DLPI error 4, unix error 233 sending to 0019bbeb9f29aa080009167e
(Only the interesting lines)
It isn't any problem on cluster, but...
No info on the forums, no info on google...
Any info about this message, please?
TIA

[Hasan Atasoy]

hi artura;

233 means there is no buffer space available.
dou you have any memory pressure in the system. it will recover itself but you need to investigate it. also applying latest arpa patches recommended.

Hasan

[Arturo Perez del Galleg]

Hi Hasan!
Yes, i see the 233 error in the errno.h but the unavailable space in the buffer is not very signifiant for me.
Which is the problem that produce this message? Not RAM problem, not signifiant network problem, not any problem...
The system's patch level is the HP recomended level.
Thanks,
Arturo

[Stephen Doud]

The error can occur intermittently, but when repeatable, system resources (kernel tuning) is a common cause of this problem.
Use GLANCE or otherwise analyze the system for memory depletion. Check the document titled "Kernel Memory Allocation at
http://docs.hp.com/en/oshpux11iv3.html

For 11.11 and 11.23, try changing the buffer cache allowance.
# kctune dbc_max_pct=10

[Arturo Perez del Galleg]

Hi all!
Thank you for yours replies!
The system's virtual memory is not very high at any moment (in the attachment the last example).
And the system load is about 0.5 (when high). At this moment:
9:12am up 160 days, 13:58, 34 users, load average: 0.11, 0.13, 0.15
And from man dbc_max_pct:
When the parameters dbc_min_pct and dbc_max_pct are
different, the size of the buffer cache grows or shrinks
dynamically depending on competing requests for system
memory.
In my sistem the values are:
dbc_max_pct 50 Default Immed
dbc_min_pct 5 Default Immed
If the parameters are dinamically assigned by the system the current value (dbc_max_pct) is only the superior limit, when the system is memory needed the value shrinks automatically (and the same for dbc_min_pct).
This is not?
Sorry my bad english.
TIA
Arturo