Runaway SMBD processes

[Christian Barth]

> I think this was covered earlier, but I missed it because I was beating
> my head against Win2K PDC support.Now I have that working, this problem
> is becoming a real pain.
> 
> At least once a day, smbd will go crazy for one or more users.It will
> create a new instance every minutes or so to serve the exact same share,
> while not killing the old instances. Everything will lock up for the
> user and I will have to shut down their machine, and restart smbd to
> clear things up.
We get exactly the same in the following unusual senario, which a 
haven't debugged any further:
Redhat 6.2 NFS-Server --> NFS-mount across the university campus --> 
Debian NFS-Client and Server of the same directory [1] --> Solaris 
2.7 NFS-Client and Samba 2.0.7-Server of the directory. On NFS-level 
everything is fine, but when using samba to copy files of this share 
randamly and not really reproducable the nt/w2k-clients hang. After a 
long time out they display a error message about a network problem. 
Using smbstatus, there is a lock on the copied file which always is 
there for some time after the copying has finished. In case of the 
failiure ther is a new smbd severing the same share to the same 
client. To clean things up, we kill just the new and the old smbd.

Christian


[1] don't ask wy we use this setup, the Debian-box is used to avoid 
having the important Solaris box hanging in case the campus network 
went down.
_______________________________________________________________________
In a world without walls and fences, who needs windows and gates? (SUN)