tftp error in client during reinstall over network

[Ryan Hurst]

Ok, I'm still trying to ignite a 10.20 client from an 11.00 ignite server. for reference, see the thread:
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0xab9ae3ed7640d71190080090279cd0f9,00.html
I got past the warning issue mentioned in the begining of that thread, but a new error has cropped up.
I made a new, clean install of HPUX 11.00 on my server system, with version B.4.1.61 ignite installed on it. This seemed to make the warnings go away when making a network recovery.
After making a network recovery archive of the 10.20 client (with no errors or warnings now), I tried to install back onto the 10.20 client. After the server rebooted the client, the client appears to try to tftp the ignite server and I get the following messages:

* Reading configuration information from server...
* tftp error log follows:
tftp> tftp> tftp> tftp> tftp> Transfer timed out.
tftp>
ERROR: Could not read /var/opt/ignite/INDEX from install tftp server.
NOTE: Retrying loadfile command...
ERROR: Failed to read "INDEX" file from install server. Checkthat the
install server's IP address is correct and the server has the
"Ignite-UX" product loaded and is available via the tftp(1) service.
Failed to startup networking, user interaction required


I pinged the client system at this point from the server system to make sure that the two systems could stil lcommunicate. I'm not sure what to do now.

[Sridhar Bhaskarla]

Hi,

First make sure your ignite server is configured for the subnet that of the client.

#instl_adm -d

Verify the IP here

Next, from another system on the same subnet, do a tftp to the ignite server.

$tftp ignite_server
tftp>get INDEX

This should be successful.

Once you verified both of them, your install should work.

-Sri

[John Dvorchak]

On the ignite server does the file /var/opt/ignite/INDEX exist? If so what are the permissions, also on the directory tree above? Also is tftp enabled in /etc/services:

tftp 69/udp # Trivial File Transfer Protocol

[Ryan Hurst]

sri:
Ok, I did as you suggested: the server and client are both on the same subnet.
I then tftp'd into the server from a third system on the same subnet. I tried the command "get INDEX" which gave me an error code 2: Access violation. I tried it again including the path this time: "get /var/opt/ignite/INDEX" ,which was successful.
John:
The file at /var/opt/ignite/INDEX does exist, and had the permissions rwxr-xr-x. I had wondered about permissions myself prior to submitting this question and had changed the permissions to rwxrwxrwx. The directory's permissions are set to r-xr-xr-x. I'll change them to rwxrwxrwx also, just in case the client needs to write to INDEX for some reason.
I checked the /etc/services file and the line "tftp 69/udp # Trivial File Transfer Protocol" was present.

[Sridhar Bhaskarla]

And what network interface are you using on the client?

-Sri

[Steven E. Protter]

you have a network problem.

Make sure the LAN card speed matches that of the switch or hub.

lanadmin -x 0

0 is the number of the interface lan0=0

You can not boot off the Ignite server unless its the built in lan card.

Have network administration hard code the switch, if thats what it is on your port to match your network settings on the Ignite client.

I'm in the process of writing up what will be a joint itrc doc with the support center that includes these problems.

If you solve the network issue, you can get to the next problem.

The image you are loading is of course a golden image for 10.20 or a make_net_recovery image? Its important.

SEP

[Ryan Hurst]

First off, sorry it took so long to reply, I was very ill yesterday.
Sri:
The lan port I am using is the one built into the motherboard of my b2000 system.
John:
Ok, lanadmin -x 0 reports back:
Nothing!
Lanscan shows the port as lan0, but lanadmin says lan0's management ID is 4. So when I type in "lanadmin -x 4" it tells me this:
100 Full-Duplex Auto-Negotiation-ON
(BTW, the built in lan interface is the only one on this client system.)
Looking at the switch that the client and server are both plugged into, I see that all of its ports are 10/100. Is this good enough, or does the switch still need to be hard coded?

[Sridhar Bhaskarla]

Hi Ryan,

Can you go into /var/opt/ignite/clients/(client_name) aand view the file config.sys?. Search for server and it should list the correct IP address of your ignite server along with netmask.

-Sri

[Ryan Hurst]

"The image you are loading is of course a golden image for 10.20 or a make_net_recovery image? Its important."

Oops, missed this one... Yes, the image I am trying to load back onto the client is a make_net_recovery image that I had just made from the client system just a little earlier.

Here again is the layout of the systems involved:
Ignite Server: 11.00 Unix + Ignite B.4.1.61
Ignite Client (my guinea pig): 10.20 + pax PHCO_20388 patch)
I am attempting to:
1) Make a network recovery archive of the client system. (successfull, no warnings or errors)
2) Perform a new install back onto the client system using the network recovery archive I created in step 1. (unsuccessfull, client can't tftp server after reboot)

[Ryan Hurst]

Ahah! Sri, you might have nailed it-
Another person was working on this ignite problem and she changed the server's IP, thinking maybe there was something else on the one we were using. I switched it back later, but it seems that config.sys has retained the temporary IP we were using. I'll correct it and see if the tftp works better. :-)

[Sridhar Bhaskarla · Accepted solution]

Ryan,

Looks like you found the problem. But the correction is not to be made to the config.sys file. It is automatically createdp from the instl_adm. So you will need to configure it using instl_adm.

For ex.,

instl_adm -d > config

Will dump the details into config. Edit it, change the ignite server IP to point to correct IP and then run

instl_adm -f config

-Sri