SUMMARY: IPC as Master Server Succeeded!!!

From: AVD Boogaart (sgbc220@sghms.lon.ac.uk)
Date: Tue Jan 12 1993 - 11:16:59 CST


Hi All!

My problems seem to have gone. After successfully installing the
Sparc IPC (thanks to you lot!!!) the problem persisted due to an
interrupt vector conflict between the PC's ethernet card and
DOS Windows. This seemed to be a rare problem as the manufacturer
suggested I'd change it as a last resort, after many frustrating
trials with other suggestions. So, the machines ping happily now
and I'm happy and thankful. Even the 0 subnet bits with a netmask
of 255.255.255.0 are OK, although 4 subnet bits and mask of
255.255.255.240 on the PC should work as well (according to one
of my `supporters'). The broadcast address can be either
192.9.200.255 or 192.9.200.0 (I've tried both).
So thank you all, especially Jan Bruun Andersen (jba@dat.ruc.dk),
who took the trouble of mailing me quite a few times and
managed to come up with some very clear ideas about the networking
and routing.

More thanks go to (see also the rest of the mail with previous
responders who helped to get my IPC master server right in the first
place!):

eckhard@de.dlr.go.ts (Eckhard Rueggeberg)
Christian Lawrence <cal@com.bellcore.soac>
ekurgpol@edu.usc.develop-law (Elmar Kurgpold)
Mike Raffety <miker@il.us.swissbank.com>
mgh@edu.harvard.bih.bihobl2 (Michael G. Harrington)
phil@edu.ohio-state.cgrg (Phil Ritzenthaler)

##################################################################

Here's my last "latest:" mail I sent

##################################################################

Hi all,

after the latest information my problems seem to have narrowed
considerately. The IPC now knows it's the master, the domainname
is OK, all yp services are running, 'ypcat hosts' matches
/etc/hosts etc. Just the PC is unreachable. For initial problems
see below, but :
PC : smis.tmr.com 192.9.200.7
IPC: arnie.tmr.com 192.9.200.6
On the Sparc IPC (SunOS 4.1.1, running NIS services):
# ping 192.9.200.6
192.9.200.6 is alive
# ping 192.9.200.7
no answer from 192.9.200.7
On the 386 PC (MS-DOS 5.0, running PC/TCP 2.05):
> ping 192.9.200.7
(positive answer in 25ms)
> ping 192.9.200.6
ping failed: Host unreachable: ARP failed

le0 is activated via (in rc.boot)
ifconfig le0 netmask + broadcast 192.9.200.255
in rc.local it says
ifconfig -a netmask + > /dev/null
(I've taken the broadcast out here)

so netmask is 255.255.255.0 and broadcast is 192.9.200.255
I understand this is a class C network (?), so does this
mean there's 0 subnet bits?
In the software on the PC, when I give 4 subnet bits, the
netmask automatically becomes 255.255.255.240
Only 0 subnet bits will give me a netmask of 255.255.255.0,
but still the IPC and PC won't ping to each other.
I'm sort of confused now. I've tried everything everyone
suggested so far, and come a long long way thanks to you, but
it seems that there's one problem left, and that's the actual
communication.

Who helps???

More thanks go to:

Jeff Mallory <jeff@com.digex.access>
root@com.irt (Jim Murff)
bill@com.access (Bill Hunter)
jba@dk.ruc (Jan Bruun Andersen) (especially helpful suggestions)
laugh@com.mot.fsd.dal (Annette D. Smith)

Original problem plus earlier update:

Hi all,

Problem persists! Message built up as:
- original question
- replies I got and what they meant to my problem
- new diagnostics (move forward, search for string 'newnew')

a few days I posted my original question:
=======================================================================
systems:
Sun SparcStation IPC (hostname: arnie)
    SunOS 4.1.1 (alas, no rev. B)
    8 MB RAM
    OpenWindows 2.0

SRS 386 (IBM-compatible) PC (hostname: smis)
    8 MB RAM
    500 MB Hard Disk
    MS-DOS 5.0
    DOS Windows 3.1
    FTP PC/TCP v2.05

We try to connect the two up in a mini-ethernet. A thin ethernet cable
connect the ethernet card in the PC with the transceiver in the ethernet
port of the IPC. The software on the PC has been installed fully.
The basic problem is, however, to make the IPC a master server.
>From the O'Reilly book "Managing NFS and NIS" I took it I needed to
run the ypinit -m command. I did the following:
edited /etc/hosts to add:
    192.9.200.6 arnie
    192.9.200.7 smis
created /etc/ethers to contain:
    8:0:20:f:22:e8 arnie
    0:0:c0:bc:de:2c smis
edited the file /etc/defaultdomain to contain the name:
    tmr.com
    rather than 'noname'
edited /etc/passwd to remove the last line which contained:
    +::0:0::
created /var/yp directory
copied NIS-Makefile:
    # cp /usr/lib/NIS.Makefile /var/yp/Makefile
tried to run ypinit:
    # /usr/etc/yp/ypinit -m
      next host to add: arnie
      next host to add: ^D
but it complained:
    /etc/netgroup: don't know how to make
    /etc/bootparams: don't know how to make
    couldn't find /etc/timezone
    couldn't find /etc/auto.master
    couldn't find /etc/auto.home
    Target 'all' not remade because of errors

I understood the last three error messages arose because I use SunOS 4.1.1
and not SunOS 4.1.1 rev. B. However, the first two didn't look too good.
I removed /var/yp again and 'touched' /etc/netgroup and /etc/bootparams.
After that ypinit only complained about the timezone etc. and ended
successfully. However, I still find it strange that the /etc/netgroup and
/etc/bootparams files are empty.

After rebooting the IPC the following error message occurred:
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
NIS: server not responding for domain "tmr.com"; still trying
etc.
In the initialization it also said:
le0 at SBus slot 0 0xc0000 pri 5
le0: RARP timed out
I can only get the IPC back to normal by rebooting to single user
mode and renaming the /etc/defaultdomain file (disables the
NIS services in the /etc/rc.local file).

So apparently the IPC somehow doesn't realize it's the master server,
even though ypinit completed 'succesfully'.

Can anyone tell me what it is that I have missed? ANY responses
are welcome, I'm quite desperate (isn't everybody who is responsible
for machines that many others depend upon) since I have evaluation
software to test on the mini-ethernet that runs out in a few weeks
time.
I'd be very thankful to those who can help me out. Of course I'll
summarize, this mailing list is of vital importance to my survival.
=======================================================================

Instantly, I got many replies, all of which were very helpful. However,
after trying all of them, the problem still persists! Here are some
replies and what they meant to my problem:

=======================================================================
From: exubho@se.ericsson.exu (Blake Holman)
make sure of uncommented lines in /etc/rc.local
#
# Master NIS server runs the XFR daemon
#
                ypxfrd; echo -n ' ypxfrd'

>> I uncommented the ypxfrd daemon, very helpful indeed, but
>> problem persists
=======================================================================
From: mgh@edu.harvard.bih.bihobl2 (Michael G. Harrington)
After running ypinit -m, you should run ypserv. This starts up the
NIS service daemons.

>> ypserv doesn't give an error message. Is also run at boottime, but
>> problem persists
=======================================================================
From: stern@COM.Sun.East.sunne (Hal Stern - NE Area Systems Engineer)
you might also be having problems binding to yourself, although
this shouldn't be an issue. to test this out, put in
        ypset IPADDR
after the ypserv and ypbind invocations, and be sure to start
ypbind with -ypsetme (edit these in /etc/rc.local). you need
to use the IPADDR of your master server since yp isn't running
yet to even convert a name to an IP address. if this works,
but removing the ypset command still causes problems, then
your NIS master is probably having problems hearing its own
broadcasts -- check netmasks and bcast addresses.

>> Wasn't recommended according to manual pages. Tried it anyway,
>> but to no avail
=======================================================================
From: kkc@edu.arizona.nursing.stat (Kee Koon Chua)
Did you forgot to uncomment the line B=-b in Makefile:
# Set the following variable to "-b" to have NIS servers use the domain name
# resolver for hosts not in the current domain.
B=-b
#B=
DIR =/etc

>> This should not be a problem, since the IPC and PC form the one domain,
there is no other domain, server or machine. Tried it anyway, to no avail.
=======================================================================
From: kubipdal@dk.uni-c.uts (Peter Dalg}rd)
One more guess: the NIS makefile gets its dommain name from a call to
domainname. I.e., the name must be set BEFORE executing make. Not
sure ypinit does this. Try
        cd /var/yp
        domainname tsr.com
        mkdir `domainname`
        make
Of course, if you already have /var/yp/tsr.com/*.{dir,pag} then your
problem obviously lies elsewhere.

>> The first was OK already, the latter seems true for me...
=======================================================================
From: Geert Jan de Groot <geertj@nl.philips.ica>
I suggest you try:
1. Make SHURE the YP database is made. check in /var/yp/tmr.com on the
   master server.
2. Check on the netmask&broadcast address of yourethernet interface
   le0 (ifconfig le0). THIS IS IMPORTANT!
3. What does domainname say?
4. With YP, try to reboot while setting up. It helps!
5. Check why your IPC complained about RARP. I can't remember seeing a
   RARP statement in the original rc.* files.

>> domainname and yp database OK, might my problem have something to do
>> with the le0? Look out for further descriptions on le0 further on in
>> this text!
=======================================================================
From: Steve J Lacey <sjl@uk.co.compnews>
I think you need to manually add in a `+' to /etc/bootparams and /etc/netgroups
- Don't forget to restart rpc.bootparamd, etc...

>> Tried it, but should not be necessary anyway
=======================================================================
From: trinkle@edu.purdue.cs (Daniel Trinkle)
     Did you set the domain name to tmr.com before doing the ypinit
command? You said you edited the /etc/defaultdomains file, but that
is only used at boot time. If you did not then set the domainname to
tmr.com, the current domainname was something else. The ypinit
command created the directory /var/yp/`domainname` and put the DBM
files in that. At boot time, it checks to see if /var/yp/`domainname`
exists. If it does, ypserv is started. However, if the domainname
command does not produce the same results at boot time as it did when
you ran ypinit, ypserv will not be started.
     So, before running ypinit, run the command
# domainname tmr.com
     Then clear out /var/yp and redo the ypinit. After doing the
ypinit, make sure the directory /var/yp/tmr.com exists and contains
lots of *.dir and *.pag files.
     The netgroup and bootparams files don't need to be there if you
don't use them. Bootparams are only necessary if you have diskless
clients that are booting from you. An empty file will keep the make
(run by ypinit) from complaining. The netgroup facility is useful,
but if you don't use it an empty /etc/netgroup file is just fine.

>> This was a relief. The rest was OK
=======================================================================
From: ats@de.gmd.first.prosun (Andreas Schulz)
Don't worry about "/etc/netgroup" and "/etc/bootparams", these are
files you need for large installations and for diskless operation.
Did you edit your "/etc/rc.local" script ? Does is start and tells
you about starting "ypserv" ?
You also said, you add the two names, wasn't the SUN in /etc/hosts
before ?

>> Yes, the SUN was already there. The rest is OK.

newnew

So, further diagnostics now are:
the yp database seems to get built up OK. The domainname is OK.
I can manually run ypserv, ypxfrd, ypbind, rpc.ypupdated. I get
messages every now and then (also at boottime):

NIS: server not responding for domain "tmr.com"; still trying

messages on le0 during boottime:
le0 at SBus slot 0 0xc00000 pri 5
le0: RARP timed out
     (I thougth one got that message when the ethernet port was empty?)
le0: flags=40<RUNNING>
        ether 8:0:20:f:22:e8

Now on a different site in our hospital there is a Sparc 10 on a mini-
net, where it is a slave server (IP address 192.1.1.3, the master server
has 192.1.1.0). Here it says:
le0: flags=63<UP,BROADCAST,NOTRAILERS,RUNNING>
        inet 192.1.1.3 netmask ff000000 broadcast 192.1.1.0
At boottime it reads:
le0 at SBus slot f 0xc00000 pri 6 (onboard)
le0: AUI Ethernet

Can anybody tell me how I might set proper values (which?) on my IPC?

I think maybe netmask and broadcast values might be changed, in the
/etc/rc.local they are put as follows:

ifconfig -a netmask + broadcast + > /dev/null

Other commands give following results:

# ping arnie
arnie is alive
# ping arnie.tmr.com
NIS: server not responding for domain "tmr.com"
# ping 192.9.200.6
sendto: Network is unreachable

WHO HELPS ME PLEASE???

I'd be very grateful!!!

Bye, Aad

-- 

+-----------------------------+---------------------------------------+ | Aad van den Boogaart | e-mail: a.boogaart@uk.ac.lon.sghms | +-----------------------------+---------------------------------------+ | St. Georges Hospital Medical School (University of London) | +-----------------------------+---------------------------------------+ | Division of Biochemistry | phone : +44 (0)81 672 9944 | | Cranmer Terrace | ext : 55809 | | London SW17 0RE | fax : +44 (0)81 784 7092 | | United Kingdom | | +-----------------------------+---------------------------------------+



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:21 CDT