On Sun, Jul 01, 2001 at 04:53:07PM +0200, Andre Alexander Bell wrote: > This is from /var/log/syslog: > > Jul 1 11:20:22 julia nmbd[2789]: connect from 192.168.76.4 > Jul 1 11:20:22 julia inetd[2757]: /usr/sbin/tcp: exit status 0x1 > ... > repeated many times. then: > Jul 1 11:20:24 julia inetd[2757]: netbios-ns/udp server failing (looping), > service terminated > 10 minutes later the same again...
I've seen this happen on some systems. It is an inetd bug afaik. Do a "killall -9 nmbd" every time you restart inetd, for a workaround. Or put it in the /etc/init.d/inetd script, as I did. > This continues for some hours every 10 minutes... > Then: > > Jul 1 14:02:42 julia kernel: Out of Memory: Killed process 2455 (apache). > even this a few times. But nothing else goes anymore... Weird, that part never occurred to me on any system. Is the machine very low on physical memory, perhaps? > Notice: > - 192.168.76.4 is the win95 compi from my neighbour. > - 192.168.76.2 is my debian/testing box. > - julia is running testing as well. > - both debian/testing boxes do have kernel 2.4.2 2.4 kernels have different swap allocation strategy. You need lots of swap with 2.4, or you may find yourself running into surprise glass walls. > - julia is configured as preferred master, local master and domain master. In that case, why don't you run smbd and nmbd standalone, instead of from inetd? That way, the inetd bug can't bite you at least. Running it from inetd is useful if you only serve any smb-related services once in a while, and if the daemons would be a waste of valuable memory that is primarily intended to be used for other tasks. > What is going wrong here. What do I have to change to get it work properly. > Can anyone advise me somehow? > (btw: in man nmbd under -D stands: running from inetd is possible but not > recommended! Why is it started from inetd in debian? Or am I wrong here?) It is started from inetd because your were asked at installation time if that is what you wanted. Apparently you did. Cheers, Joost