RADs and inetd

[Mark Poeschl_2]

We are getting ready to put some new GS1280s (8 CPUs in each) into production and I have a question about the inetd 'child' processes I see on the system. Excerpt from the inetd man page:

"-r radid
Specifies the identifier of the Resource Affinity Domain (RAD) on which to start an inetd child daemon. You can specify this option multiple times on the command line (see the "Examples" section). The default is to start a child daemon on all RADs."

I'm only seeing child inetd processes for the first four (lowest numbered) RADs. 'vmstat -R' shows all 8 RADs - numbered 0, 1, 4, 5, 8, 9, 12, and 13; but there are child inetd processes only for RADs 0, 1, 4, and 5.

Is this the expected behavior? Is there something I need to configure to let inetd know about the other for RADs?

[Michael Schulte zur Sur]

Hi,

have anyone changed /sbin/init.d/inetd ?

greetings,

Michael

[Mark Poeschl_2]

Certainly not intentionally. This is for Tru64 5.1B with Patch Kit 5 applied:

epic_prod:backup# ls -l /sbin/init.d/inetd
-rwxr-xr-- 1 bin bin 2587 Apr 14 2003 /sbin/init.d/inetd*

The 'start' portion of the script makes it look like the inetd binary is responsible for firing up its children.

[Ann Majeske · Accepted solution]

This is a known problem with inetd. If you want the patch you should open a problem report with HP Tru64 UNIX support.

Ann Majeske

[Vladimir Fabecic]

Ann,
You said this is a known problem with inetd.
So, why should someone request a patch? HP should make it without special request! Am I wrong?

[Ann Majeske]

The patch will be in the next patch kit. If you want it before the next patch kit, you should open a problem report with HP Tru64 UNIX support.

Ann

[Mark Poeschl_2]

Got the early-release patch and it solved the problem