Posts made by chad-bisd

Right now, the inventory task ends with a clearing of the task file and a reboot. If you want to perform an inventory and then do more stuff, I think you’ll have to modify that behavior without interfering with the rest of the tasks that call the inventory scripts.

I’m sorry, but I’m confused. Does the image upload ok, but die during the deploy, or is it dying during the actual upload? Just trying to narrow down the window of when the error occurs.

Does it actually fail and die at this point, because that error message indicates a problem with the USB ports of the client.

You have to have a machine (virtual or otherwise) already loaded up with Ubuntu, take that image, then deploy that image to another machine (virtual or otherwise). You can’t easily use Fog to push the Ubuntu installer. Also, if you are going to image Linux, you can only have primary partitions of type ext2 or ext3. Fog does not currently support ext4 filesystem, which is the default for Ubuntu. When you install Ubuntu, specify the advanced options during partition and formatting to use only ext2 or ext3 for your non Swap partitions.

In vmware, I think you have to tweak the network adapter settings to make it so the two virtual machines can talk to each other. In VirtualBox, you have to make them both part of the same named network. I do not know the specifics for vmWare.

Here: This will accept all pending MACs without question. This is a hack, and not an elegant way. It does try to wrap the statements in a transaction to avoid leaving macs in the pending table that have been approved, or removed macs that didn’t get added successfully.

If someone wants to add some MySQL error handling to this, feel free. It won’t hurt my feelings.

[CODE]BEGIN;

INSERT INTO hostMAC (hmHostID, hmMAC) SELECT pmHostID, pmAddress FROM pendingMACS;

DELETE FROM pendingMACS;

COMMIT;[/CODE]

I use phpmyadmin, and run this as a bookmarked query in the fog database.

You can also save this as a .sql file and use run it against your database from the command line:

[CODE]user@fogserver:~$ mysql -u root -p[password] fog < approvependingmacs.sql[/CODE]

Maybe even schedule a cron job to automatically approve pending macs on a schedule.

sounds like you have a conflicting configuration with your original switch. Jumbo Packets, duplex mode, or something wonky?

I found the link to the thread talking about the 3.4 kernels the links in the thread pointing to filedropper don’t work. they go to the filedropper home page, not to an actual file. Maybe the files have expired from filedropper?

The latest I can get out of sourceforge is 3.2.4 and the latest off the kernel download pages on sourceforge is 3.3.3

Did you also change the image definition associated with each host?

go ahead and do:
[CODE]user@servername:~$ sudo apt-get update
[/CODE]
followed by:
[CODE]user@servername:~$ sudo apt-get upgrade[/CODE]

Then reinstall fog over itself using the same settings you did the first time. Any data you have created should remain intact, but some new packages may get installed or existing packages reconfigured.

I don’t know why you’re getting that tftp error, but something isn’t correct in your tftp setup.

Did you try a 3.4 kernel? I think that one has support for this card, which I believe uses the e1000e driver in Linux.

I reread your question and it seems you aren’t even getting the “loading kernel bzImage…” text. Is that correct?

When you installed Ubuntu, did you update and upgrade before you ran the installfog.sh?

Ask your administrator about turning on portfast for all the ports so there no delay between them being connected and being active. See if that helps.

I’m not sure why your dhcp daemon is not autostarting. Did you try adding it to the default runlevel or making sure the startup script is getting call? dmesg or /var/log/syslog might have some clues.

Are you getting a message on the client about the queue being full and having to wait for a slot to open up? I’m looking at the screenshots, the first one with the warnings during startup looks normal. The second one looks like it’s waiting on a queue slot to open up on the Fog server.

Did you enable portfast on the switch or give the units 30 seconds of wire connectivity before trying to boot them? Might be a STP issue with the ports not being active by the time the clients try to PXE boot.

What version of Ubuntu are you using and did you reboot the Fog server between imaging attempts?

What version of Fog on which version of Ubuntu?

Is this a standard “(N)ormal” install?

What brand/model of clients are you trying to PXE boot?

What is the static IP address of the Fog server interface you are using?

What version of Ubuntu are you using for your server? It sounds like you may be using 12.04 because of the tftpd-hpa issues, which is a common problem with 12.04 and documented on these forums.

I hope your deploy goes well with the different kernel.

If my fog server is inside virtualbox along with the test clients, it works fine for me. I have not yet tried to register a virtualbox guest from a fog server that is outside the host. I don’t know how that works.