[cifs-protocol] Re: CAR - missing documentation on *SMBSERVER in SMB and SMB2

Steve French smfrench at gmail.com
Thu Jun 5 18:06:00 GMT 2008


Trace now attached (forgot to attach it to the initial CAR request)

On Thu, Jun 5, 2008 at 1:03 PM, Steve French <smfrench at gmail.com> wrote:
> In trying to understand whether Windows Server 2008 allows SMB2
> connections over port 139 (due to firewalls for example on 445), I
> noticed that it is not documented whether SMB2 connections over port
> 445 require first sending a RFC1001 SessionInitialize over the socket
> (as SMB/CIFS does).   Our experiments indicated that port 139 requests
> require the SessionInitalize request, but also showed that connections
> to *SMBSERVER failed (*SMBSERVER is the default called name used by
> modern Microsoft clients) with "Negative Session Response Called Name
> Not Present" (see frames 47 and 48 of attached trace).   It appears
> that the behavior of this has changed in Windows Server 2008 which
> requires that the user now specify the target system's netbios name in
> the session initialize (which did work see frames 8 and 9 of the
> attached trace).
>
> Attached is a trace of two mounts to Windows Server 2008 Enterprise -
> the first a mount which worked and specified the servernetbiosname in
> RFC1001 session init (followed by an umount tree disconnect, ulogoff
> of that smb session which is not of interest here) and the second
> mount which failed (see frames 47 and 48) which specified "*SMBSERVER"
> as the called name and failed.
>
> I would like to see documentation of SMB and SMB2 documents updated to
> reflect when the use of *SMBSERVER is permitted and any other
> differences in use of 139 (e.g. if SMB2 has additional limitations
> when run over port 139 instead of 445).
> --
> Thanks,
>
> Steve
>



-- 
Thanks,

Steve


More information about the cifs-protocol mailing list