[prev in list] [next in list] [prev in thread] [next in thread]
List: samba
Subject: SAMBA digest 318
From: samba () anu ! edu ! au
Date: 1995-01-26 7:10:20
[Download RAW message or body]
SAMBA Digest 318
Topics covered in this issue include:
1) Re: DHCP patches
by elf@netcom.com (Marc Singer)
2) DHCP, rev
by elf@netcom.com (Marc Singer)
3) Re: Performance
by peb@pine.dk (Per Baekgaard)
4) transparent LF <-> CR/LF mapping with Samba ?
by Peter Funk <pf@artcom0.north.de>
5) RE: WFW chat (Unix talk)
by Joerg Schlaeger <joergs@q.toppoint.de>
6) LPQ on AIX 3.2.5 with Samba 1.9.00
by ggoodey@sblack.demon.co.uk (Graham Goodey)
7) warning about solaris
by "Andrew.Tridgell" <tridge@cs.anu.edu.au>
8) Workgroup login
by Bret Giddings *NOT AUTHENTICATED* <bret@essex.ac.uk>
9) RE: SAMBA digest 317
by billp@se.tandem.com
10) RE: SAMBA digest 316:new WfW PCs cannot see samba servers
by billp@se.tandem.com
----------------------------------------------------------------------
Date: Wed, 25 Jan 1995 00:30:46 -0800 (PST)
From: elf@netcom.com (Marc Singer)
To: A.J.Cole@cbl.leeds.ac.uk (Andrew J Cole)
Cc: samba@cscgpo.anu.edu.au (Samba)
Subject: Re: DHCP patches
Message-ID: <199501250830.AAA16895@netcom9.netcom.com>
> Have you made any progress on this? You will undoubtably have fixed the
> overlength MAC address for deloris which was causing the "bad addr len".
The funky packet that caused this trouble was not comming from
deloris. About an hour after I sent that message, I discovered that
it was a recurring event on the net. Pauline pointed out that the
first three bytes are the letters RAS, Remote Access Service. It
appears to be something to do with Microsoft's service, but I cannot
stop it from occuring.
> I too am unclear what Pauline means by a broadcast address of
> 255.255.255.255. MS TCP/IP assumes the standard network braodcast
> address of all 1s which is why Sun shops (like ours) use the lmhosts
> file on the Samba server to ensure that the server appears in the
> Workgroups using an all 1s braodcast.
What do you mean here about the lmhosts file. We have a Windows NT
server that is used only as a data server. My Linux machine is
hosting all of the DNS and Samba services. Everyone is now using DHCP
to setup their TCP/IP addresses and DNS to resolve names into network
numbers.
The trouble I was having was that even though the DHCP packets were
arriving at the Linux machine, it was unable to send them back due to
the way that my routing tables were configured. I has a default route
that was pointing to a PPP device. With this changed to the eth0
device, all is better. I have toyed with ipconfig (on the PC) a bit
and noticed that there is no DNS server being loaded via DHCP. It
seems to work OK because the DNS server is the default gateway. I
have considered trying to use Samba as a WINS server which is supposed
to work with version 1.9.01.
> I havent had a chance to setup a test network and switch a bootp host
> to using all 1's as its own broadcast address yet and havent noticed
> anyone else reporting success.
This last bit confuses me to. I have played with adding a route for
255.255.0.0, but it never gets used. I would be most appreciative if
you could explain how to get my Linux machine to route packets for
address 255.255.255.255 onto the ethernet while retaining a default
route for the ppp device.
--
Marc Singer
elf@netcom.com
------------------------------
Date: Wed, 25 Jan 1995 00:36:55 -0800 (PST)
From: elf@netcom.com (Marc Singer)
To: iay@threel.co.uk (Ian A Young)
Cc: samba@cscgpo.anu.edu.au (Samba)
Subject: DHCP, rev
Message-ID: <199501250836.AAA17383@netcom9.netcom.com>
>
> > This is the text of a message I sent to Pauline directly regarding my
> > experience with bootp/dhcp. I expect that either a) I am not setting
> > the botptab file correctly, or b) my Windows client is not properly
> > configured. Has anyone had positive luck?
>
> We've got to this stage as well. Did you get past it? Did you get a reply
> from Pauline? Even a supposedly-working bootptab file, or a packet dump
> from a working NT DHCP transaction would be a great help.
I found that there were three key pieces. First, the bootptab must have
the correct MAC address and syntax for the target machines. You can
test that the bootptab records are correct by sending a USR1 signal to
bootp and looking at the /tmp/bootp.dump file. DHCP does not really
require a special tab entry, but there may be some extra fields that
will make it configure more completely.
Second, check that the request packets are comming in and being
matched. Use the -d4 switch on the bootp command line to test this.
Look in the /var/adm/messages file (or cognate) for the bootp entries.
Last, if the requests match but the target host does not configure
properly, you probably have a routing table that does not route the
response packets out onto the same net where the request was received.
Pauline has stated that the address 255.255.255.255 must be routable
back onto the ethernet, so make sure that your default route will do
this. She has suggested that another route for network 255.255.0.0
can be added to teh routing tables, but I was unable to get this to
work.
If you still want tables, send me another message and I will mail them
to you.
--
Marc Singer
elf@netcom.com
------------------------------
Date: Wed, 25 Jan 1995 10:44:52 +0100 (MET)
From: peb@pine.dk (Per Baekgaard)
To: samba@anu.edu.au
Subject: Re: Performance
Message-ID: <m0rX4HU-0003iQC@pine.pine.dk>
<HANNU@estib.ee> wrote:
> WfW3.11 client: 486DX2/66, 8MB
> netcard DLink De250 (ISA), IDE Disk (ISA), 8MB mem
> Linux server (1.0.9), 386SX/16, 5MB
> netcard DLink, Samba 1.9.00, IDE Disk (ISA)
> SUN SparcServer 5 (Solaris 2.3), network and disk whatever it came
> with, Samba 1.8.05
>
>
> Server Linux SUN
> ----------------------------
> Write speed 98 70
> Read speed 98 444
>
For what it is worth, here are my figures:
WfW3.11 client 486DX/33MHz, 16 MB mem / wfw-tcpip32-a
netcard 3-com 3c509 (Elink III) (ISA), IDE Disk (ISA)
Axil Sparc-2 clone running Solaris 2.3, 32 MB mem
3 SCSI disks in total (Micropolis 0.45/1.3/1.7 GB)
My timing figures are the 'raw' speeds, i.e. when reading from the
server I read into the null device using large blocks (dd if=f:/blabla
of=nul bs=32k). Using smaller blocks (i.e. using copy instead of the MKS
dd) does affect performance, but not much (10%). Writing data to local
disk instead of to nul make it run approx. half speed, memory serving me
[Note: I've since installed NT, and has given away the disk with WfW
(now reformatted), so I cannot repeat the test with WfW].
Read speed 715 kB/s
Write speed 692 kB/s
Combined speed 350 kB/s [ read _and_ write on net ]
Reading data of the local disk is slightly slower!
Doing the same test under NT shows similar performance (i.e. around 700 kB/s
read/write, depending on what the server is doing otherwise), and reading of
a local disk is also around 700 kB/s -- but combining net read and local
write makes if much faster because the two now gets interleaved (around
500 kB/s) and/or I don't use dblspace on the disk.
Thus, the network card in use seems to affect the performance tremendously!
For comparison, I've done a 'real world' test under wfw. When backing up
my local disk (270 MB/dblspace compressed) I could do so in half an hour
using XCOPY. It used to last for 2-3 hours using PC-NFS against the same
server...
-- Per.
--
> Per Baekgaard, Pine Tree Systems peb@pine.dk :
> Koegevej 62, 1., DK-2630 Taastrup, Denmark :
> Phone: (+45) 43710704 Fax: (+45) 43714342 :
------------------------------
Date: Wed, 25 Jan 1995 12:58:50 +0100 (MEZ)
From: Peter Funk <pf@artcom0.north.de>
To: samba@cscgpo.anu.edu.au
Subject: transparent LF <-> CR/LF mapping with Samba ?
Message-ID: <m0rX6N8-0003QeC@artcom0>
Hello all !
I've recently installed 'samba-1.8.05' on several SCO Unix and Linux
besed computers and stumbled over the following question : I would
like to make some unix-Textfiles available on a PC running WFW 3.11.
Since DOS/WFW uses CR/LF as EOL, it is necessary, to convert between
LF and CR/LF, when the data is exchanged with WFW-client.
After studying the supplied files samba.faq, *.5 *.8 and the source
code of server.c and reply.c I have come up with the conclusion, that
this is impossible without hacking the source and extending the
functionality of 'read_file', 'write_file' and 'seek_file' seriously.
I think, that a table of EOL-positions will be needed to accomplish
such a task. Am I right ?
Please reply directly, since I haven't joined this mailing list.
(at least not now)
Regards, Peter
-=-=-
Peter Funk, Oldenburger Str.86, D-27777 Ganderkesee, Germany
office: +49 421 2041921 (ArtCom GmbH, Grazer Str.8, D-28359 Bremen)
------------------------------
Date: Wed, 25 Jan 1995 14:52:36 +0100 (MET)
From: Joerg Schlaeger <joergs@q.toppoint.de>
To: samba@anu.edu.au
Subject: RE: WFW chat (Unix talk)
Message-ID: <199501251352.OAA00573@q.toppoint.de>
Hi,
I send some code to Andrew to implement the
chat from WFW to samba.
(using SMBsends, SMBsendstrt, SMBsendend, SMBsendtxt)
I use this for sending messages from the WFW stations to the
postmaster of the unix host.
Joerg
--
Joerg Schlaeger joergs@toppoint.de
------------------------------------------
(to be faster with the /2)
------------------------------
Date: Wed, 25 Jan 95 11:09:07
From: ggoodey@sblack.demon.co.uk (Graham Goodey)
To: samba@cscgpo.anu.edu.au
Subject: LPQ on AIX 3.2.5 with Samba 1.9.00
Message-ID: <3012@sblack.demon.co.uk>
Hello all,
I am using Samba 1.9.00 with AIX 3.2.5, WfWg with MSTCPIP v3.11a
I have recently upgraded from 1.8.05 where I was using the lpq script
which was posted on comp.protocols.smb.
SInce upgrading samba 1.9 I have been unable to interogate the print
queue. I have removed the "print command" and "lpq command" entries in
smb.conf and added "printing = aix".
Has anyone had similar problems.
Printing works correctly.
Many thanks in advance
Graham
--
Graham Goodey
Internet: ggoodey@sblack.demon.co.uk
------------------------------
Date: Thu, 26 Jan 1995 01:25:45 +1100
From: "Andrew.Tridgell" <tridge@cs.anu.edu.au>
To: samba@cscgpo.anu.edu.au
Subject: warning about solaris
Message-ID: <199501251425.BAA04907@arvidsjaur.anu.edu.au>
If you use solaris 2 then be very careful not to compile with the
"bsd" compiler. It's often in /usr/ucb/cc.
If you do then you will get some very strange errors. In particular,
directory listings will be totally stuffed up.
It seems that the problem is that it will link with bsd libraries that
have a different format for "struct dirent". This means things like
"dptr->d_name" will come out garbled.
It took quite a while to track this problem down with one person :-)
Andrew
------------------------------
Date: Wed, 25 Jan 1995 16:16:07 -0500 (EST)
From: Bret Giddings *NOT AUTHENTICATED* <bret@essex.ac.uk>
To: samba@anu.edu.au
Subject: Workgroup login
Message-ID: <ECS.9501251607.L@servelan.essex.ac.uk>
Is it possible to use samba as the password server for a given Workgroup domain. If \
not, why not. Is it outside the scope of the SMB spec and hence samba?
Just curious - but a positive answer would make my day/month (maybe even year!)
Bret
--
Bret Giddings, Systems Programmer, Computing Service, University of Essex
Tel: (0206) 873585 Email: bret@essex.ac.uk Fax: (0206) 860585
------------------------------
Date: Wed, 25 Jan 95 08:38:46 PST
From: billp@se.tandem.com
To: samba@anu.edu.au
Subject: RE: SAMBA digest 317
Message-ID: <Chameleon.4.01.950125084656.billp@bubbles>
> Hi. I am running Samba 1.9.01 on both FreeBSD 2.0R and SunOS 4.1.3
> using WfWG3.11/MS-TCP and NT3.5 clients.
>
> On both sets of clients, I can see the FreeBSD samba server in
> the browse list, but I cannot see the SunOS server.
>
> On the NT machine, I have to type "NET USE X: \\host\share" from
> the command prompt, and cannot get anything using FileManager.
>
> On the WfWG machine, I can type "\\host" in FileManager
> and it will list the shares, but the "host" listing dissappears
> as soon as the connection is made.
Browsing didn't work for us on SunOS 4.1.3 until we added the -G
option to the nmbd run line. There also a couple of params you
might want to check in smb.conf.
*********************************
* Name: Bill Prince *
* E-mail: billp@se.tandem.com *
*********************************
------------------------------
Date: Wed, 25 Jan 95 08:52:59 PST
From: billp@se.tandem.com
To: samba@anu.edu.au
Subject: RE: SAMBA digest 316:new WfW PCs cannot see samba servers
Message-ID: <Chameleon.4.01.950125085641.billp@bubbles>
> I have not loaded any TCP/IP stuff onto the PCs. Question: do PCs have to
> have TCP/IP running to access the samba servers? I was under the impression
> that samba would work with just the WfW protocol, NetBEUI.
Whoops! You absolutely, positively _have_ to have TCP/IP installed on
the PCs in order to use Samba. In fact, I would go so far as to say
to make it the primary protocol. Also note that run-of-the-mill TCP/IP
packages don't support NBT (NetBIOS over TCP/IP), so make sure that
yours does. The good news is that the MS TCP/IP includes NBT, and it's
free.
bp
*********************************
* Name: Bill Prince *
* E-mail: billp@se.tandem.com *
*********************************
------------------------------
End of SAMBA Digest 318
***********************
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic