[Samba] setup_new_vc_session error message on samba-3.4.8/solaris10

[Joe Cammisa]

folks, we're running samba-3.4.8 (--with-krb5=/usr/local/lib --with-ads
--with-ldap) under solaris_10; in the past week we've seen an issue where
the number of smbd processes begins to far exceed the number of
connections (as reported by smbstatus).  when this occurs, we see the
server in question stop answering new requests (ssh, smb, nfs) though old
requests are honored and system logs report nothing unusual.  stopping the
samba process with pkill does not kill all of the smbd processes (which
must be done manually); afterwards, samba can be started but still new
requests are not noticed until we resort to a reboot.

one other item of note--i see lots of these sorts errors in log.smbd in
the time leading up to the interruption:
[2010/09/20 13:00:40,  2] smbd/sesssetup.c:1360(setup_new_vc_session)
  setup_new_vc_session: New VC == 0, if NT4.x compatible we would close
all old resources.

not sure what to do about this--perhaps using ndd to allow for more tcp
requests?  anyone else see such problems before and have any advice? 
thanks in advance for your attention.  best...

-joe




our config is pretty basic (as follows) and the number of simultaneous
connections ranges between 30 and 70:

[global]
workgroup = WORKGROUP
netbios name = BLABLA
server string = STORAGE
bind interfaces only = True
interfaces = bge0
deadtime = 20
debug level = 2
security = user
password level = 8
encrypt passwords = yes
socket options = TCP_NODELAY
follow symlinks = yes
wide links = yes
unix extensions = no