ghost machines, revisited

Jim Watt wattjg at appliedbiosystems.com
Thu Jan 3 14:10:07 GMT 2002


On Thu, 3 Jan 2002, Keith Warno wrote:

} OK I think I've got it cleaned up.
}
} On each of the sambs servers I nuked all files in /var/cache/samba/ and
} restarted the daemons, starting with the WINS box.
}
} It appears the problem was incorrect usage of "remote browse sync" by
} some of the samba master browsers.  They were configured to sync with
} the WINS server and being all the SMB boxes on this context are in the
} same workgroup, this "remote browse sync" was unecessary.  Somehow the
} ghost machines would linger around in one of the boxes' browse lists and
} then get forcefully synced up with the WINS server, thus tainting the
} lists everywhere.
}
} This is only a theory.  :)

Looks like cranking up debug level to 4 for nmbd will produce *lots*
of output from a function in nmbd_packets.c called debug_browse_data().
How much of it would be useful/revealing I don't know. :)

I discovered there's a handy utility called "smbcontrol" for setting
debug levels these days.  It replaces SIGUSR1/SIGUSR2's functionality,
among other things, judging from the manpage.  Looks kind of useful!
I should read the manpages oftener! :-)

Jim
-- 
Jim Watt                               wattjg at appliedbiosystems.com
Applied Biosystems                     Voice (desk): +1 408 577 2228
3833 North First Street                Fax:          +1 408 894 9307
San Jose CA 95134-1701                 Voice (main): +1 408 577 2200





More information about the samba mailing list