Solaris x86 Jumpstart pxe boot = TFTP ERROR - Access Violation

Look in the directory /tftpboot on gibson. You should see a file there called "01001EC9B11AEE" that is a symbolic link to the real bootfile, which should also reside in the same directory. Is that the case?  If it is, check that the real boot file has world readable permissions.

Also I was looking at a friends jumpstart server and he has the hex files listed for the clients he adds. Is that my problem? How do I generate those files.

Also this may help here is my addclient command with the output, doesn't seem like its making that hex macro file or whatever it is:

# sudo ./add_install_client -e 00:1E:C9:B1:1A:EE -s 10.212.52.250:/data/install/b109 -c 10.212.52.250:/data/jumpstart -p 10.212.52.250:/data/jumpstart i86pcump

updating /etc/bootparams
ln: cannot create /tftpboot//pxegrub.I86PC.Solaris_11-1: File exists

Create a grub floppy and edit GRUB menu to contain
the following entry:
min_mem64 1024
title Solaris netinstall
  rarp
  kernel$ /I86PC.Solaris_11-1/platform/i86pc/kernel/$ISADIR/unix -B install_media=ip-addr-for-10.212.52.250:/data/install/b109,install_boot=10.212.52.250:/data/install/b109/boot
  module$ /I86PC.Solaris_11-1/$ISADIR/x86.miniroot

Yes, that is the problem. The hex file links should be there. And yes, it would appear that once the problem
is seen by the add_install_client, it doesn't finish the work.

What you might try to do is empty the whole /tfptboot file, and start over, checking after you run the add_install_client to see if everything looks right. I suspect that it got an error somewhere and now can't recover.

Assuming your clients IP is 10.212.52.250 just create a symbolic link in /tftpboot on your boot server:

cd /tftpboot
ln -s pgegrub.I86PC.Solaris_11-1 0AD43455

and kick off jumpstart again.

Blu; unmounted the tftp mount and deleted the /tftpboot directory and I re-ran the addclient...still got the same problem.

Rowley: The client ip is 10.212.52.85, but the hex concertion between 10.212.52.250 and 10.212.52.85 is the same...that did not work. I also tried to make the sym link above but with the hex file it was looking for originally 01001EC9B11AEE and that did not work either....same issue. I am really at a loss here.

Well, that certainly won't work, since Rowley left off the ethernet type "01" at the begining. However, he has a point. I thought of just having you try that, but if there is one problem, there might be others and it is probably cleaner to start over.

blu: what do you mean by start over? I deleted the /tftpboot directory and everything....what else can i do?

Can you post your /etc/ethers and /etc/hosts config, as well as an 'ls -l' of /tftpboot ?

Rowley:

  Here is the `ls -l` of /tftpboot after i did the add_client command:

ls -l /tftpboot
total 296
drwxr-xr-x   3 root     root         512 Mar 26 10:52 boot
drwxrwxrwx   5 root     root         512 Mar 24 13:56 I86PC.Solaris_11-1
lrwxrwxrwx   1 root     root          26 Mar 26 10:52 nbp. -> pxegrub.I86PC.Solaris_11-1
-rwxr-xr-x   1 root     root      138384 Mar 26 10:52 pxegrub.I86PC.Solaris_11-1
-rw-r--r--   1 root     root         124 Mar 26 10:52 rm.10.212.52.85

Here is my /etc/ethers:

00:1E:C9:B1:1A:EE  testjump

and here is the entry in the /etc/hosts:

10.212.52.85    testjump

any thoughts? ALL the documentation i look at it states that at the bottom of the output after you run the addclient command it should tell you the boot file it made...which is what the PXE server is saying it cannot find because it cannot fine it in /tftpboot, in confident once we figure out why add_client is not creating this 01[MACADDRESS} file it will fix my issue.

Let me know what else you need

Okay, so you deleted the /tftpboot directory and re-ran add_install_client and still didn't get the link made? Was there any error messages the first time you rant it?

Okay, try this.

Delete the /tftpboot directory.

Run the script command.

While inside the script shell, run the add_install-clinet script, but feed it to the shell:

sh -xv add_install_client -e ....

where .... is the rest of the command line.

Once it is finished, exit the script shell and post the typscript file here.

blu: you totally lost me.....run what script command?  Whil,e inside what script shell? the add_install_client script runs and thats it, i  cant do anything while "inside" it

I'm confused

The "script" command. See man script(1).

The script command creates a sub-shell session with all of the characters written to a file called "typescript".  This file will have the (huge)
debug output from running the add_install_client command.

The typescript file is attached, thanks i never knew about the script command. Let me know what else you guys need

Okay, do this:

Check the /etc/mnttab file to see if there are any lofs mounts of /data/install/b109 or its sub-directories onto /tftpboot or any sub-directories.
If there are, use the umount command to unmount them.
Then edit the /etc/vfstab file and remove any lines that create lofs mounts of /data/install/b109 onto /tftpboot or subdirectories.

After that, delete the /tfpboot directory.

Double check that everything in /data/install/b109/boot looks like it is okay. If you ran a "rm -rf" on /tftpboot while there were
lofs mounts, it might have deleted something in there. Maybe recreated those locations from the disk if possible. Specifically, look
for a file /data/install/b109/boot/multiboot.

After all of that, do the "script" thing again. If it works, great. If it doesn't, post the new typescript file you got.

I did everything you asked, no go...same issue..../data/install/b109/boot/multiboot is there and everything appears as it should,same size as on cd as in the jumpstart media directory. Attached is the typescript file for this run...i think it will prob be the same exact as last one. But here it is anyway. Any other ideas?

You're missing your symlink to the tftp grub pxegrub file -check the code snippet - you're totally missing that. You NEED a hex symbolic link pointing to the driver. For x86 jumpstarts there's some additional files, bits and bobs i've never had experience with as I only have sparc systems. This guy has a written a decent looking howto.

http://solaris.reys.net/english/archives/63-Solaris-PXEDHCP.html

Good luck.

bash# ls -al /tftpboot/0100093D0068F3
lrwxrwxrwx   1 root     other         28 May 23 17:27 \
/tftpboot/0100093D0068F3 -> pxegrub.I86PC.Solaris_10.1-1

                                              

1:
2:
3:

Rowley:

I tried your suggestion, same issue.

Here is the ls -al of the /tftpboot directory i manually created that sym link:

drwxr-xr-x   5 root     root         512 Mar 26 13:20 .
drwxr-xr-x  37 root     root        1024 Mar 26 12:52 ..
lrwxrwxrwx   1 root     root          36 Mar 26 13:20 01001EC9B11AEE -> /tftpboot/pxegrub.I86PC.Solaris_11-1
drwxr-xr-x   3 root     root         512 Mar 26 13:13 boot
dr-xr-xr-x   5 nickv    unixadm      512 Feb 23 07:09 I86PC.Solaris_11-1
drwxrwxrwx   5 root     root         512 Mar 24 13:56 I86PC.Solaris_11-2
lrwxrwxrwx   1 root     root          26 Mar 26 13:13 nbp. -> pxegrub.I86PC.Solaris_11-1
-rwxr-xr-x   1 root     root      138384 Mar 26 12:52 pxegrub.I86PC.Solaris_11-1
-rw-r--r--   1 root     root         124 Mar 26 13:13 rm.10.212.52.85

I still get the TFTP - Access Violation error on the client

ok ok...i think i got it, i followed this guide:

http://solaris.reys.net/english/archives/63-Solaris-PXEDHCP.html

and it would appear i have to run the following command in x86 for it to get the boot file made, i ran it and it generated the file.


./add_install_client -d -e 00:1E:C9:B1:1A:EE -c 10.212.52.250:/data/jumpstart -p 10.212.52.250:/data/jumpstart -s 10.212.52.250:/data/install/b109 -b "console=ttya" i86pc

now...my server jumped...BUT when its doing the install process after the splash screen where the dots are moving..............................it stalls about half way in the 2nd line. Any ideas?

Actually, that got rid of one error I saw, about the mount point . Now I see the problem.

You using add_install_client without the "-d", which means to use RARP. But it appears that PXE only works with DHCP, not RARP. You
need to set up the DHCP and follow the instructions for DHCP installs. Too bad add_install_client doesn't say anything about this. Apparently, you can use a floppy disk and install from there, so it cannot just assume that if you are using PXE you must use DHCP.

So, to answer the last message, you must set up a DHCP server and add the macros mentioned in the add_install_client output.

blu.....this is the reason why my installing is halting on the 2nd line afte rthe solaris_jumpstart splash screen it gets to the 2nd line of ................................................................

??

I think so. If the system is getting the no macro or the wrong ones, I would expect it to hang at some point. Try the snoop again to see what
is going on.

Blu, i dont think that is it as when i modified my add_client_install command it created the correct boot files.........i dont see anything in snoop that would point to an issue...i see it doing tftp and nfs back and forth and its all ok....and then it just stops.

blu also when i tried your suggestion with my old add_client_install script using the dhcp manager utility to add the boot files its looking for it told me they already exist.

Hmm. You are apparently getting the boot files okay, and all of the installation files. Could it be that it is actually starting the install and you just don't see progress on the screen? Does the snoop stop showing traffic?  If so, what is the last thing it shows?

Here is a sniplet during snoop, it does all this and then just stops:

      gibson -> testjump     TFTP Data block 50225 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50225
      gibson -> testjump     TFTP Data block 50226 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50226
      gibson -> testjump     TFTP Data block 50227 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50227
      gibson -> testjump     TFTP Data block 50228 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50228
      gibson -> testjump     TFTP Data block 50229 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50229
      gibson -> testjump     TFTP Data block 50230 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50230
      gibson -> testjump     TFTP Data block 50231 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50231
      gibson -> testjump     TFTP Data block 50232 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50232
      gibson -> testjump     TFTP Data block 50233 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50233
      gibson -> testjump     TFTP Data block 50234 (1432 bytes)
    testjump -> gibson       TFTP Ack  block 50234
      gibson -> testjump     TFTP Data block 50235

any thoughts? Still stalling.....since yesterday I wiped EVERYTHING out..dhcp server, jumpstart configs, and tried new media..i started from scratch and followed this guide

http://brandonhutchinson.com/wiki/PXE_JumpStart

still stalling at the 2nd line of the .................................................. with the same as above when I do the snoop. Any thoughts?

From the snoop, can you tell what what file is being downloaded? It would have been earlier in the snoop.