NetBoot service not starting clients (10.6 Server)

[dayzd]

Hoping someone can help shed some light on a netboot problem I'm having.

 

I'm in the process of setting up my 10.6.4 server and cannot get any clients to boot from my netboot image (made using DeployStudio).

 

The bootimage was made on my newest hardware (iMac10,1, OSX 10.6.4) and I've since tried a bootimage made on my server and my previous generation iMac (9,1), none of which successfully boot. (I know the image from my 9,1 iMacs works as I used it last year before the current redeployment process was started.)

 

I've seen the '10.6 If your netboot clients won't start' (Server Admin 10.6 Help: If NetBoot Client Computers Won?t Start) document, and this has offered no insight into this issue.

 

I've also seen the more in depth 10.3/4 document about netboot clients not starting (Mac OS X Server: NetBoot clients cannot start up from server (NetBoot troubleshooting)). Is the process still similar in 10.6? I can see all the events listed, except the tftp START entry in my logs.

 

Below are the NetBoot logs when I try to start a client with the 'N' key at boot:

 

Sep 20 09:43:28 clw-sr-xserve1 bootpd[208]: BSDP DISCOVER [bond0] 1,0:25:4b:b0:11:ba NetBoot003 arch=i386 sysid=iMac9,1
Sep 20 09:43:28 clw-sr-xserve1 bootpd[208]: DHCP DISCOVER [bond0]: 1,0:25:4b:b0:11:ba
Sep 20 09:43:28 clw-sr-xserve1 bootpd[208]: replying to 192.168.15.102
Sep 20 09:43:28 clw-sr-xserve1 bootpd[208]: OFFER sent clw-wk-imac-002 192.168.15.102 pktsize 300
Sep 20 09:43:30 clw-sr-xserve1 bootpd[208]: NetBoot: [1,0:25:4b:b0:11:ba] BSDP ACK[list] sent 192.168.15.102 pktsize 310
Sep 20 09:43:30 clw-sr-xserve1 bootpd[208]: DHCP INFORM [bond0]: 1,0:25:4b:b0:11:ba
Sep 20 09:43:30 clw-sr-xserve1 bootpd[208]: reply ciaddr 192.168.15.102
Sep 20 09:43:30 clw-sr-xserve1 bootpd[208]: ACK sent clw-wk-imac-002 192.168.15.102 pktsize 300
Sep 20 09:43:31 clw-sr-xserve1 bootpd[208]: BSDP INFORM [bond0] 1,0:25:4b:b0:11:ba NetBoot003 arch=i386 sysid=iMac9,1
Sep 20 09:43:31 clw-sr-xserve1 bootpd[208]: NetBoot: [1,0:25:4b:b0:11:ba] BSDP ACK[list] sent 192.168.15.102 pktsize 310
Sep 20 09:43:31 clw-sr-xserve1 bootpd[208]: DHCP INFORM [bond0]: 1,0:25:4b:b0:11:ba
Sep 20 09:43:31 clw-sr-xserve1 bootpd[208]: reply ciaddr 192.168.15.102
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: BSDP INFORM [bond0] 1,0:25:4b:b0:11:ba NetBoot003 arch=i386 sysid=iMac9,1
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: replyfile /private/tftpboot/NetBoot/NetBootSP0/NetBoot-DS-Runtime-01Sept10.nbi/i386/booter
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: NetBoot: [1,0:25:4b:b0:11:ba] BSDP ACK[sELECT] sent 192.168.15.102 pktsize 416
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: DHCP INFORM [bond0]: 1,0:25:4b:b0:11:ba
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: reply ciaddr 192.168.15.102
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: ACK sent clw-wk-imac-002 192.168.15.102 pktsize 300
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: DHCP DISCOVER [bond0]: 1,0:25:4b:b0:11:ba
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: replying using broadcast IP address
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: replying to 255.255.255.255
Sep 20 09:43:37 clw-sr-xserve1 bootpd[208]: OFFER sent clw-wk-imac-002 192.168.15.102 pktsize 308
Sep 20 09:43:45 clw-sr-xserve1 bootpd[208]: DHCP DISCOVER [bond0]: 1,0:25:4b:b0:11:ba
Sep 20 09:43:45 clw-sr-xserve1 bootpd[208]: replying using broadcast IP address
Sep 20 09:43:45 clw-sr-xserve1 bootpd[208]: replying to 255.255.255.255
Sep 20 09:43:45 clw-sr-xserve1 bootpd[208]: OFFER sent clw-wk-imac-002 192.168.15.102 pktsize 308

 

Can anyone identify what might be going wrong?

The XServe has a bonded uplink to the switch, each client has a gigabit connection. All are Intel machines.

 

(Question also asked on the Apple Discussion forums: Apple - Support - Discussions - NetBoot service not starting clients ...)

[HodgeHi]

Are these brand new models? Are you using the same .nbi file to boot the newer machines as you are using to boot the older ones? If so this may be the cause. You may need to create a newer netboot set than the one you are using. The graphics frameworks may have changed causing the issue. The graphics framework is used to display the GUI for the clients.

 

Also have you started a booting process for a machine and then forced quit it at any point. If so, try restarting both the deploystudio service (in System Prefs) and also the NFS service and netboot service. I have once seen an issue where after manually shutting down a machine hard during a restore the netboot service then refused to boot any more clients until i restarted the services.

 

Also have you got the model filter enabled? if so and these are new iMac models has the server admin tools been updated to have the newer models in the filters list? Have these new machines been added to the filter list?

Edited September 20, 2010 by HodgeHi
Added last question.

[dayzd]

Are these brand new models? Are you using the same .nbi file to boot the newer machines as you are using to boot the older ones? If so this may be the cause. You may need to create a newer netboot set than the one you are using. The graphics frameworks may have changed causing the issue. The graphics framework is used to display the GUI for the clients.

 

Also have you started a booting process for a machine and then forced quit it at any point. If so, try restarting both the deploystudio service (in System Prefs) and also the NFS service and netboot service. I have once seen an issue where after manually shutting down a machine hard during a restore the netboot service then refused to boot any more clients until i restarted the services.

 

Also have you got the model filter enabled? if so and these are new iMac models has the server admin tools been updated to have the newer models in the filters list? Have these new machines been added to the filter list?

 

While I do have newer iMacs than I had last year, as I stated above, I have remade my nbi on my newest hardware and it does not boot either model. (I ran an older iMac with the the old nbi as a test, which also failed.)

 

I've not forced-quit any netboot processes, as so far, none of my clients have started from it. The server has also been restarted a couple of times while I've been trying to sort this out, so I can't see it being that (although I'll try again!).

 

Model filtering is not switched on. Interestingly, on last years (9,1) nbi, if I do turn it on, there are model identifiers available in the list, but for the new nbi (from the 10,1 iMac) there are not...

[HodgeHi]

what version is your server admin tools? I think it is upgraded when the server is upgraded but not always.

 

Also if you are working from a client machine you will need to update the tools on there as well.

[dayzd]

I'm using 10.6.4 on the server and for the admin tools.

[HodgeHi]

have you seen this site (originally from Bombich)

 

Bombich.com: Troubleshooting the NetBoot Process

[dayzd]

Just had a quick scan of that link. Looks like it could well be helpful!

 

Would've checked there sooner, but we've had problems with our internet connection since the summer and the LEA had been very restrictive with their filters... Must be sorted now, though!

 

Will check it out tomorrow and let you know (can't access the room for the rest of the day).

 

Thanks for all your help.

[HodgeHi]

No problem

[dayzd]

Well, I had trouble getting any definitive answers yesterday...

Myself and my boss (who knows very little about Macs, but is a smart man) went through that afp548 netboot page and found it difficult to identify which scenario applied to us: some things were true for each one, and some things weren't!

We tried all sorts and sadly got nowhere...

 

I read a few new links last night, some suggesting that the OSX Server firewall ought to be off, others that said it should be on (with appropriate ports open) and one rather discouraging post stated that the only fix for the (very similar sounding) problem they were having was to re-install OSX Server (which, for time reasons, isn't really an option)!

 

My next step is to do more firewall fiddling and to try and rule out the switch (we've turned off spanning tree on the end points to help with connection speed).

Failing that, I've got a G4 box with 10.4 server on at home, which I'm tempted to bring in and test, just to rule out cabling/network issues!

 

Oh, I hate this problem!

[HodgeHi]

I am rather limited to the amount I can help as all forums are restricted here except this on e and a couple others. It's quite annoying. I did find a forum post that, i think in the apple discussions before when I had a problem with Netboot.

 

It was to do with checking the NFS mount points.

I would disable the firewall to do some checks to ensure that the firewall is not causing the problem. You can always enable it again afterwards.

 

Also check your DeployStudio version. It may be an issue of incompatible versions.

 

You could also try moving the NetbootSP0 and NetBootClient0 shares. It maybe that the permissions have gotten messed up somehow. Doing this I think re-creates the shares with appropriate permissions.

 

Also make sure that AFP is working correctly.

 

I will see if I can find the info I used for my problem but I wouldn't hold my breath that I can get to it!

 

You could even try putting the boot up in verbose mode by holding down cmd +V