Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.hawaii.edu!news.uoregon.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!torn!nott!nntp.igs.net!usenet
From: cskinner@bml.ca (Chris K. Skinner)
Newsgroups: comp.unix.bsd.freebsd.misc,list.freebsd-questions,local.freebsd.questions
Subject: Re: routed timing out my LAN card so tcp/ip seems not to work.
Date: Tue, 28 May 1996 15:21:09 GMT
Organization: Bytown Marine Limited, Nepean/Kanata, Ont, Canada
Lines: 386
Message-ID: <4of5sk$mtn@nntp.igs.net>
References: <4o7vfb$m0h@nntp.igs.net> <31A9DBF5.446B9B3D@FreeBSD.org>
NNTP-Posting-Host: ttya13.ott.igs.net
X-Newsreader: Forte Free Agent 1.0.82
At 09:44 AM 5/27/96 -0700,
"Jordan K. Hubbard, Walnut Creek CDROM,
President, FreeBSD Project"
<jkh@FreeBSD.org> wrote:
>Chris K. Skinner wrote:
>> FreeBSD was said to fit into 700 Mbytes, but this did
>> not work with the default label program setups. This
>I need a lot more information than this - what do you mean by "default
>label program setups" and "did not work" in this context? I routinely
>install into 200MB and 300MB partitions without any difficulty
>whatsoever, to say nothing of larger ones.
[big
snip]
Just what I said:
1. Get a Seagate ST32140A 2.1 Gbyte EIDE hard disk.
2. During install, tell FreeBSD to make 2 partitions
on the clean drive. The first one should be 440 Mbytes.
The second should be the rest. The first one, although
marked as FreeBSD is just a quick means to put a 440
Mbyte Gap in to hold a MS-Dos/MS-Windows partition.
3. Request that the boot manager for FreeBSD be installed.
4. Delete the first 440 Mbyte partition so that that
portion can be claimed for use with MS-Dos/MS-Windows
at some later stage.
5. Insure that the remaining FreeBSD partition is marked
as Active/Bootable.
6. Save and exit from the fdisk utility.
7. Now the labeling program (whatever it is called) begins.
Select the option that creates default labeling for
the lone-FreeBSD partition.
Save it and Bingo!
At this stage, the software has selected labeled portion sizes
for the various UNIX file system parts that "are going to
run into trouble later on in the install because one (or
maybe two) of the labeled portions are a little on the
small size, and fill to overflowing during the
'Everything' install."
The only stuff of the Ports/Packages
that I left out of things was anything having to do with
the Japanese character set support--anything that mentioned
Japanese in the description.
As I said, I couldn't recall if it was /var or what that
was too small.
I chose /tmp as the place where stuff got loaded temporarily.
I'm sure that at some points during the install
that the /var portion got filled, and then
after the error was reported and the installer
moved on to the next component for installation,
and it removed the build-up clutter from the overflow
that happend, and that's why I smashed the ant with
a steam roller by changing var from 30 Mbytes
to 300 Mbytes. You'll see that it's now only 1%
full...
This is the current state of my file system:
===========================================
bash# df
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/wd0a 60430 58736 -3142 106% /
/dev/wd0s1 449800 215088 234712 48% /mnt
/dev/wd0s2f 1125118 726068 309040 70% /usr
/dev/wd0s2e 302222 3706 274338 1% /var
procfs 4 4 0 100% /proc
/dev/cd0a 650002 650002 0 100% /cdrom
/dev/wd1s2 45206 44808 398 99% /mnt2
/dev/wd1s3 482568 481536 1032 100% /mnt3
/dev/wd2s1 482568 355480 127088 74% /mnt4
/dev/wd2s2 45206 2 45204 0% /mnt5
bash#
=============================================
Each of the file systems mounted on /mnt /mnt2 /mnt3
and so on are all Dos/Windows partitions on other
attached EIDE hard disks that were already housed
in the UNIX box, but now are wired up to run as
primary slave EIDE, secondary master EIDE, and
secondary slave EIDE disks, and a couple of them
have "Ontrack Disk Manager" MS-DOS file system
'extended partitions' that are positioned beyond the
MS-DOS/FAT 1024 cylinder limit, but are mountable
by FreeBSD's mount_msdos command (praise be to
you FreeBSD Gods!).
As you can see, my / (root) file system seems to be
running at 106%--this can't be good?--Is it?
What do you suggest that I do at this point?--Backup
a bunch of my hand-crafted configuration files to
the networked DOS machine and go for another install?
I dunno! I'm just a Gumby when it comes to this
UNIX stuff! Please be patient with me--a dummy
MS-Dos/Windows guy--so that I can be weaned from
the curse of MS-Dos/Windows and join the
UNIX / X-windows club.
>a) Figuring out whether this was pilot error or a genuine bug.
I realize that I've got to crack some books on the
various topics, but every keyword mentioned in
e-mail responses and reply news postings is more
meaningful to me if the language is long, plain
English, and pertains to the rough spots that I've
encountered. These anecdotes are probably somewhat
useful to others looking on who see the advice and
help that has been offered to me. In addition, as
a developer of sorts myself, I realize that it is
sometimes tough to see the forest for the trees and
get a different perspective of the work being
undertaken. If I'm just a distraction, then so
be it, but I believe that my gripes and complaining
are based on a viewpoint that may have a large
constituency--MS-Dos/MS-Windows weenies--who are
trying to come to grips with the neat stuff that you
and the FreeBSD gang have created.
>I'm sorry, but a lot of this message simply comes across (whether or not
>you so intended it) as mere bitching and whining, not as a genuine
>attempt to educate or help in the solution of the problem.
Don't be offended or upset by criticisms. If some of
these gripes and complaining can be addressed
by some measures like easier installation+configuration
of the software and slightly broadened FAQ or handbook info,
then FreeBSD and other unices will have a larger
support base of talented users and developers.
If it isn't--it should be an unwritten goal of
developers to gain converts away from the Microsoft camp?
What better way to do this, than by addressing the
griping and criticisms in the positive ways that you
have been: requesting clarifications on items of
possible concern, rectifying any legitimate shortcomings,
and carrying on the good fight! It only adds to your
stress levels to get emotionally charged by some
mistaken, off-handed remarks that I and others make.
If I've gotten out of hand, I apologize. I wine a lot!
>Finally, since I mentioned personal style, might I perhaps persuade you
>to shorten your posts a little? We both would appear to share a common
>failing of not using 10 words when 100 would do just as well, and one of
>us is already one too many. :-)
The reason that my postings are quite long is that I'm
not sure what details to leave out.
Some people when they post a news article asking
for help to get their system to go leave out
so much that no one wants to react to
their posts--the literate few wait for someone else
to tell them,
"Hey, tell me how big your disk is; What kinda disk
controller are ya using?; How much Ram d'ya got installed?;
what hardware checks did ya do on the cards and internal
cabling?; What drive geometry did ya use?; Did ya try to
get support from your computer dealer?; ... and other
such."
I try to cram all the little phrasings in that might
possibly be relevant so that others might tweak onto
one of them that may point to a trouble spot, and
give me suggestions on what to try out next.
I got the following advice from
"Doug J. Santry" <dsantry@maccs.dcss.mcmaster>, an interested
news article reader:
>When it asks you how you want to slice your partition, make only 2 slices,
>multiply the amount of RAM you have by 2, make this your swap, and the rest
>of your disk stick it under /, then /var and /use will go under / automatically
>and you won't have that problem anymore.
He had pointers about books to read, the differences in
X-windows/UNIX users culture versus MS-Dos/MS-windows users
culture, and configuration tips about religiously
backing up crucial config files that might get squashed
during re-installs of the software.
Other readers noticed that I mentioned that I was having
troubles with getting X-windows apps like Netscape
to run properly because of improper config and
setup, so they pointed me out to a bunch of
helpful examples and live configuration files and
notes: "ftp://pm.cse.rmit.edu.au/210m" and
"ftp://pm.cse.rmit.edu.au/210l", he and others pointed
to several windows managers that might be worth trying.
At 09:44 AM 5/27/96 -0700, "Jordan K. Hubbard, Walnut Creek CDROM,
President, FreeBSD Project" <jkh@FreeBSD.org> continued writing:
>I really think this one was pilot error - I've installed Samba during
>the install dozens of times and it's worked just great! If it did not
>come up "out of the box" for me I'd have continued working on the samba
>configuration portion of sysinstall until it did! As it stands, you're
>the only one who's ever complained.
Samba:
I've read a bunch of news articles from people
who've nearly got the thing to go right, but not quite.
Now this is me too! I followed the man pages for
smbd, smb.conf, nmbd, inetd.conf, services, and such
but my UNIX box seems to only use
/usr/local/samba/bin/smbclient \\\\cskinner\\c -N
properly because the WFWG 3.11 host, "\\\\cskinner,"
has a valid IP because in addition to Netbeui it is
running the MS-TCP/IP-32 stack for its LAN card.
The following were the settings:
1. The /etc/services file already had the suggested
entries therein for the operation of samba:
netbios-ns 137/tcp #NETBIOS Name Service
netbios-ns 137/udp #NETBIOS Name Service
netbios-dgm 138/tcp #NETBIOS Datagram Service
netbios-dgm 138/udp #NETBIOS Datagram Service
netbios-ssn 139/tcp #NETBIOS Session Service
netbios-ssn 139/udp #NETBIOS Session Service
2. The /etc/inetd.conf file did not have any of the
man page suggested entries that corresponded to
smb or samba or nmbd, so I added some
(each of a & b on one config line):
(a)
netbios-ssn stream tcp nowait root
/usr/local/samba/bin/smbd -d3
-s /usr/local/samba/smb.conf
-l /usr/local/samba/slog
(b)
netbios-ns dgram udp wait
root /usr/local/samba/bin/nmbd
-M "-" -d3 -T15 -S -n UNIX
-H /usr/local/samba/hosts
-l /usr/local/samba/nlog
3. Two smb.conf files could be found in the system:
/stand/smb.conf and /usr/X11R6/src/smb.conf
>Again, I'm going to write this one off as pilot error. I don't know
>*what* you did, but you clearly did something which cockroached the
>Novice install's samba configurator (and if you didn't use the Novice
>install then you really really should have and I have no sympathy
>whatsoever for you :-).
I've done as little typing as possible, during
this whole thing (excepting these news article
postings) and do not remember plunking these files
down where they were found, nor creating one in
a random place just for fun. I've been using the
Novice install each of the 5 times through.
I used vi and examined and edited the /stand/smb.conf
file. It seemed to have the options that I had
chosen during installation, so I augmented it a bit
while flipping Alt-F1/Alt-F2 console pages to different
login sessions where I had the man 5 smb.conf pages
on the topic displayed on screen. I came up with
the following file that mostly clarified
some the defaults by explicitly spelling them out:
======================================
[global]
comment = FreeBSD - Samba %v
log file = /var/log/samba.log
dont descend = /dev,/proc,/root,/stand
printing = bsd
load printers = yes
lock directory = /tmp
locking = yes
strict locking = yes
map archive = yes
dead time = 0
admin users = chris cskinner ckskinner
status = yes
public = yes
read only = no
default case = lower
short preserve case = yes
preserve case = yes
strip dot = no
security = share
guest ok = yes
null passwords = yes
force group = WORKGROUP
force user = no
workgroup = WORKGROUP
getwd cache = yes
hide dot files = yes
keep alive = 60
auto services =
[homes]
browseable = yes
comment = User Home Directory
create mode = 0775
public = no
writeable = yes
[printers]
path = /var/spool
comment = Printers
create mode = 0700
browseable = no
printable = yes
read only = yes
public = no
[unixroot]
path = /
comment = the root of UNIX filesystem
read only = no
browsable = yes
create mode = 0775
public = yes
writeable = yes
======================================
I copied the file to the /usr/local/samba directory
since it did not have anything else in there except
the bin sub-directory.
The /usr/local/samba/bin/smbclient software was able to
list all of the browsable shares on the \\cskinner
MS-Windows host and perform other operations like
"more DBLSPACE.INI", "ls" and other miracles, but
it could not do the same on itself, even when given the
specific TCP/IP address of itself. It came back with
"non-specific error: return code 0." (or some such),
"host computer was uncooperative" or similar wording.
When browsing was attempted from the \\cskinner MS-WFWG 3.11
computer, no "UNIX" host computer name or its shares could
be found. When WFWG 3.11 was given the exact name of the
computer and a share, "\\UNIX\UNIXROOT", in any combo of
upper/lower case mixture that would seem sensible, the
returned message from MS-WFWG 3.11 was "Error: Network Busy".
No password file has been setup for samba yet--Will it be
necessary--It shouldn't be now, but...?
The -H /usr/local/samba/hosts file that is used for my nmbd
command line has the following content:
=============================================
# this is the /usr/local/samba/hosts file
0.0.0.0 WORKGROUP G # set local host's Group
cskinner.bml.ca CSKINNER S # set browsable domain
hamshack.bml.ca HAMSHACK S
alexpc.bml.ca ALEX'PC S
tonymason.bml.ca TONYMASON S
john.bml.ca JOHN S
workshop.bml.ca WORKSHOP S
=============================================
The #-comments on the side of a couple of lines
above were only added here to clarify and do
not exist in the actual file.
This was supposed to identify the machine's group
as WORKGROUP and the network names of the browsable
machines on the LAN. Only the CSKINNER machine,
with its installed MS-TCP/IP-32 Beta 1.1 (?) software,
could even be manually contacted or detected by
the samba client shell software. When the other
listed machines had their NetBeui names entered
in-place on the same command line that had
functioned correctly for the \\cskinner machine
name, the operation just hung there until ctrl-c
was pressed to abort smbclient.
I invite all readers to identify any goofy errors that
they see with the stuff that I've configured--I've
probably screwed up something again...
Any help would be appreciated. Regards, Chris K. Skinner.