nmbd quitting/looping

[Terry Lambert]

> 
>         I have recompiled (1.9.17p2) and moved the samba software from a
> SparcStation 20 running Solaris 2.4 to a Netra NFS running Solaris 2.5.1 and
> are suddenly experiencing a problem where the nmb daemon is quitting with
> the following syslog entry:
> 
> Oct  6 11:17:41 lindwall inetd[158]: netbios-ns/udp server failing
> (looping), service terminated
> Oct  6 11:28:59 lindwall inetd[158]: /usr/local/samba/bin/nmbd: Child Status
> Changed

[ ... ]

> But, unfortunately I don't have the facility to modify & recompile the SUN
> Solaris 2.5.1 inetd. I am interested as to why the change in architecture &
> OS is suddenly causing this problem.
> 
> Has anyone experienced (and fixed?) this problem? 

-----------------------------------
If the problem is not what you say:
-----------------------------------
Is it possible that you have configured a udp service as "nowait" in
inetd.conf?  This is incorrect usage.  See "man inetd.conf" on your
Solaris box.

-----------------------------------
If the problem is what you say:
-----------------------------------
Each vendor's inetd has what it considered "reasonable" values for
the number of connections to an inetdmanaged port in a given interval.
They also have a minimal time threshold for how long they have to run
before they are considered to have died from unnatural causes.

The problem you are experiencing is relative to the amount of traffic
to a prticular port in a given time interval, and the average lifetime
of the daemon when that happens.

In other words, the Netra's idea of "reasonable" is different from
that of the Sun release of the OS.

I believe you can get around this problem by running the daemon
as a standalone program instead of running it from inetd (gross,
but functional).


					Terry Lambert
					terry at cs.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.