[Samba] Samba server, works fine for several days, then load increases indefinately till server unavailable

[James A. Dinkel]

Our Samba server running on CentOS 5.1 (upgraded from 5.0) with Samba
3.0.25b-1.el5_1.4 has been in place for about a year and a half.  We had
no issues till about 2 months ago.  This may have coincided with the
upgrade of CentOS 5.0 to 5.1 (which would have brought a new Samba
version with it).

 

Anyway, the server will be fine and snappy for a week or so, then out of
the blue, nobody can connect.  Top shows a few smbd processes maxing out
the cpu and the load (which is usually < 1.0) gradually climbs up to 10,
20, and onward.  If I stop smb and winbind with the init scripts, then
the load goes through the roof.  I've waited to see if it will clear
itself up, but the load skyrockets up over 1000 so I kill all the smb
processes with "sudo kill -9 `pgrep smbd`" and then the load stops
climbing and after about 5 or 10 minutes will be back down to under 1.0.
At that point I can restart smb and winbind and everything is ok again.

 

I don't see anything unusual with the logs.  Can anyone help?

 

James