RIS Stopped Working Last

[Brian Anderson]

Help!!!

Last November I configured the 1 server in my environment to begin a RIS
deployment. The server already had AD, DHCP and DNS configured and had
been in operation for 15 months. I added RIS, inserted both the default
CD image and an image created specifically for the PCs in this office.
There are 20 identical PCs here (HP eVectra, 550Mhz, 256MB RAM, 9GB HD,
LanWorks-3Com NIC, PXE 2.02). I began the deployment and everything
worked perfectly for the first 18 systems. I did a handful of systems a
month and had finished system #18 on January 15th.

Last Thursday 2/12, I patched the server in response to the security
bulletins that appeared. Since then I haven't been able to get RIS to
work. The 2 clients I have left to finish both sit at the screen trying
to acquire a DHCP address and then respond with "No DHCP or boot offers
received." After this the network boot process is terminated. I have
checked every log I know of on the server and nothing is being logged
regarding this. Just to verify that everything was working for DHCP I
added a brand new system to the network and got an address immediately.
I have rebooted the server a bunch of times, run the RIS verification,
and triple checked the server's DHCP authorization. I even uninstalled
the patches I had applied last week which forced me through a reboot but
didn't fix the problem. Reinstalled the patches (another reboot) and I
am still having the same problem. Since 1/15 when I used RIS to
successfully set up 3 systems I have made no changes to the server other
than the patches last Thursday.

Any ideas?

TIA,
Bandy

 

[Brian Anderson]

A follow on to my original message.

I noted that the BINLSVC is now generating a 1003 event in the
Application log nearly once a minute. In reviewing older logs it seems
that these event used to happen every 4 hours or so. I don't know if
this is a symptom or if one of the patches might have altered the way
this service works but it definitely changed last week.

 

[NIC Student]

We use those same eVectras in our environment. I think I understand that
your new machine worked ok and you RIS'd it but the old ones still won't
work.

Before we make changes on your server, do you have a USB floppy that you can
plug into the e-Vectas and use the 3Com boot floppy. This will tell us
whether the PXE rom needs to be updated on the eVectras. The 3Com disk
works better than the Windows RBFG floppy generated one.

Go to my site and click on the 3Com boot floppy link:

http://www.mvps.org/serverstuff/

 

[Brian Anderson]

I did not RIS my system since it didn't get a response either but I was able
to get a DHCP address, use email, surf the web etc. All regular functions
seem to be running normally. I have tried 2 other system types today and
none of them are able to get a connection to the RIS server. In one of
those cases it had previously been RISed on another network.

I am not using a boot floppy for these installations. I have configured
them to boot to the network and the firsst 18 worked like a charm.

 

[NIC Student]

Can you RIS on the same subnet as the RIS server?

Look in the DHCP scope and see if the RIS clients are getting an address.

Now if the clients are getting an address, see if that ip is already in use
on the network by pinging.

Check the authorization and the "Respond to.." and "Do not Respond.." boxes
are checked/unchecked according to this article:

PXE Client Receives a "No Reply From Server" Error During RIS Server
Discovery
http://support.microsoft.com/default.aspx?scid=kb;en-us;284987

Try starting a RIS on one of the pcs that was previously RIS'd. Don't RIS
it all the way, just see if it will start the process correctly.

--
Scott Baldridge
Windows Server MVP, MCSE


"Brian Anderson"

 

[Brian Anderson]

There is only a single subnet with the 20 workstations on it. I should
also note that all 18 of the systems I successfully configured were
connected to the same cable and data jack. I RIS 2 systems, do final
config on each system for a specific user and then swap the new system
in place of the user's existing PC. Just in case there is some sort of
problem I even tried RIS from a different data jack with no success.
And I still don't understand why the BINLSVC present 1003 informational
events every 60 seconds.

This is pretty frustrating.

 

[NIC Student]

Let me know the results these items:

 

[Brian Anderson]

- I do not see the RIS clients getting a DHCP address.
- The first box is checked and the second is not checked
- I've verified all settings as noted in that KB article and I still
cannot get the RIS clients to acquire a DHCP address

 

[NIC Student]

Well, if you are not getting a DHCP address, there is something more going
on than a RIS problem. We need to find out why the machines are not getting
DHCP first. You said your have checked that DHCP is authorized. Have you
unauthorized, then reauthorized the DHCP server?

Is this server multi-homed? Make sure the server is authorized by ip
address rather than by computer name.

Do you have enough addresses in the DHCP scope to allow for more leases?

Are you sure the nics are functional in the two workstations?

Do you have a USB floppy to put on the machines so they can boot from floppy
instead of the PXE nic for these tests?

What is the exact error 1003 message you are receiving?

--
Scott Baldridge
Windows Server MVP, MCSE


"Brian Anderson"

 

[Brian Anderson]

I just tried this again with a new system and got the same errors. On
the workstation that I am trying to RIS I get an error which says

"PXE-E51 No BOOTP or DHCP offers received."

Some other notes:
- I have 2 other laptops from vendors that were brought in to this
network for the first time today and they both received DHCP addresses.
- I have already Unauthorized and Re-authorized DHCP. No Change
- I have 50 addresses set aside for DHCP and only 23 in use. Just for
kicks I added 20 more to the scope with no change.
- I took a couple of the systems that had been previously configured
using RIS and tried to re-run the RIS install and got the same error.
- PXE boot disk using the USB floppy got the same error.

 

[NIC Student]

OK, let's make sure of:

You have run Start>Run>RISETUP -CHECK>OK on your RIS Server and it says
there are no problems.

You can delete the DHCP lease for a client, then after attempting to RIS,
you can see a lease has not been given.

Your RIS server has only one nic. *important*

RIS, DHCP and clients are on one subnet.

You have applied Service Pack 4 to the RIS Server.

Try booting from the 3Com boot floppy instead of the Windows one, get it
from here:
http://www.mvps.org/serverstuff/

Do you have spanning tree turned on your switch/router between the clients
and RIS? Pause your clients by hitting the pause button on the keyboard as
it is searching for DHCP. Wait 20 seconds and hit any key to resume. Let
me know if spanning tree is enabled.

You can email the system and app logs from the RIS server to me at
(e-mail address removed) (remove NOSPAM).

--
Scott Baldridge
Windows Server MVP, MCSE


"Brian Anderson"

 

[anonymous]

To fix this, I disabled 'Spanning-Tree' (STP) on the switches (I was using
HP Procurve/Cisco Catalyst)

This causes a delay in network connection when the system boots up long
enough for DHCP to timeout.


HTH

 

[Brian Anderson]

I'll answer all of this is the same order you presented it.

Yes - ran the RISETUP -CHECK and is verified that everything was all right.
Even ran it from the GUI in case it did something different and it also said
the RIS implementation was fine.

Yes - if I delete a lease and then try to RIS it fails. If I don't delete
the lease and try to RIS it fails. When I boot normally the system
connects, refreshes the lease and works normally.

Yes - the server only has one NIC.

Yes - the server had SP4 applied. I also applied all other patches
including the patch for the ASN1 problem This is the point at which the RIS
communications stopped working. I tried backing that out but no luck.

Yes - I used the 3Com floppy at the link you list. I had discovered this
link as a recommendation in another thread.

No spanning tree - it is a Cisco switch which has been set universally to
portfast. And it should be noted that the port in use haven't changed
anywhere along the line. It also doesn't matter which port I use.

I could forward you the log files but there is nothing to see. I never get
a message on the server except for the standard 1003 BINL started
successfully messages. All I get are the PXE-E51 "No BOOTP or DHCP offers
received" messages which occur only on the client. It appears that the
clients never see the server.

I have had to convert 1 of the last 2 systems manually because it had to be
deployed. It is possible (even likely) that the last system will also have
to be deployed in the next couple weeks. If this happens it will render the
RIS configuration useless. I have a feeling I will never know what happened
to RIS after 18+ successful deployments.

 

[anonymous]

Sounds to me like you have a network problem as opposed to a server setup
problem. This is why you can get an IP from DHCP when you boot up Windows
normally.

Have you tried running DHCP, RIS and Client in an unintelligent switch or
hub?

 

[NIC Student]

Brian, this is a frustrating one. I still am thinking it's the Cisco that
is the problem.

Can you try one more test:

Plug the Risserver into a hub.
Plug your client into the same hub.
Plug the hub into the usual switch to which the RIS server is normally
connected.
Check connectivity from RISServer>hub>rest of network.
Test RIS with the 3Com disk, F12 on startup , and the windows RBFG boot
floppy.

Let me know.

--
Scott Baldridge
Windows Server MVP, MCSE


"Brian Anderson"

 

[NIC Student]

Hi Brian,

For the test, let's also try and get the DHCP/RIS/Client all attached to the
hub so DHCP requests won't be timed out by the Cisco if that's the problem.
Did you try pausing the RIS client for 20 secs as I suggested earlier?

--
Scott Baldridge
Windows Server MVP, MCSE


"NIC Student" <