System Hangs - Boot from SAN device

dholtz-docbox

Server

FOG Version: 1.3.0-RC11
OS: Ubuntu 14.04

Description

When this system boots into PXE and has no tasks, it hangs after prompting, “Boot from SAN device 0X80.” I have read several articles related to resolving this, and they have not been very successful.

For starts, here is the output of: fog/service/ipxe/boot.php?mac=…&arch=x86_64…

#!ipxe
set fog-ip 10.1.10.42
set fog-webroot fog
set boot-url http://${fog-ip}/${fog-webroot}
sanboot --no-describe --drive 0x80

I have configured everything to boot with EXIT, so I am not sure why it is still trying to boot with SANBOOT either.

I have…

Configured everything to boot with EXIT, but the host seems unaffected
Checked the SATA configuration - it is set to IDE
Checked the BIOS for SECURE BOOT - there is no option for SECURE BOOT

Another item to note, the system has two drives. Would drive ordering matter? Only one drive has a bootable OS, and has no issue booting without going through PXE.

This will probably be my next test, changing the drive boot order - we boot off the second drive, for this system.

edit: The boot order is irrelevant. I am also currently reading through: https://forums.fogproject.org/topic/5607/stuck-uploading-image-booting-from-san-device0x80 - it has been the most relevant so far.

-Dustin

dholtz-docbox

Okay… I may have solved this at the end of the day, finally.

I believe it was a sequence of a few items not matching up once I tested certain phases. In the end, I had to switch the boot priority so that the drive with the OS had priority over the data drive. I tried this earlier, but it was after already moving past SANBOOT and when looking into EXIT as an option instead. Now that I have corrected the drive and changed the boot priority, SANBOOT is functioning as intended. I will do a few more tests before I officially resolve this, so I will post back by tomorrow morning on my results - time is limited today.

-Dustin

dholtz-docbox

So I took a look at what service/ipxe/boot.php did, and had a question/concern looking at lib/fog/bootmenu.php. In the constructor, _exitTypes is defined as…

self::$_exitTypes = array(
            'sanboot' => $sanboot,
            'grub' => $grub['basic'],
            'grub_first_hdd' => $grub['basic'],
            'grub_first_cdrom' => $grub['1cd'],
            'grub_first_found_windows' => $grub['1fw'],
            'refind_efi' => $refind,
);

Shouldn’t this also include the ‘exit’ type? I was trying to determine why it ignored my request for the ‘exit’ type in lieu of ‘sanboot’, and wasn’t sure if this was why just yet. I am still reading through the code to better understand the process though.

-Dustin

Tom Elliott

Yes it should.

dholtz-docbox

Did it get removed at some point? Looking at the GIT trunk reflects this as well: https://github.com/FOGProject/fogproject/blob/2718a13d2bd11d4d9ccd4be7f2f005a67000da3e/packages/web/lib/fog/bootmenu.class.php. What should it be updated to reflect? Because exit is not present in the array, I believe it hits @ l.339 …

        if (!$exit || !in_array($exit, array_keys(self::$_exitTypes))) {
            $exit = 'sanboot';
        }

… and defaults to sanboot instead of the selected exit type.

-Dustin

dholtz-docbox

I am going to perform a test, defaulting it to ‘exit’ in the event that the type is unknown. I will be right back.

edit: Well, that certainly eliminated the hangup, but the system fails during chain-loading now and enters a reboot loop.

-Dustin

Tom Elliott

@dholtz-docbox I corrected this for the current “working-RC-12” branch in git.

You should have this all fixed if you run:

wget -O /var/www/fog/lib/fog/bootmenu.class.php https://raw.githubusercontent.com/FOGProject/fogproject/0c8cf54f35f694504af9af5e9fcd525d2521ae60/packages/web/lib/fog/bootmenu.class.php

dholtz-docbox

Oh, awesome! Let me give it a whirl!

-Dustin

dholtz-docbox

First, we aren’t quite there yet. Second, things have progressed I believe.

boot.php now returns…

#!ipxe
set fog-ip 10.1.10.42
set fog-webroot fog
set boot-url http://${fog-ip}/${fog-webroot}
exit

However, there is still a chain-loading issue. I am unsure if this is in the boot class or not - I assume it is, given its proximity. That said, I am not sure at what step it has a chain-loading issue yet.

edit: I am still using RC11, should I try pulling down the dev-RC12 branch in its entirety and applying it?

-Dustin

dholtz-docbox

I see where it chains into boot.php in the /tftpboot directory, but am fumbling around determining what it is trying to chain into next still. I feel like this is the end of its chain, but it doesn’t realize it - when using exit.

edit: I stumbled across the following document and am trying it: https://wiki.fogproject.org/wiki/index.php?title=Boot_looping_and_Chainloading

-Dustin

Tom Elliott

@dholtz-docbox Based on what I can see, you’re running in “no-menu” mode?

dholtz-docbox

@Tom-Elliott : Correct.

-Dustin

Tom Elliott

@dholtz-docbox So when is it getting the “chainload” error?

dholtz-docbox

@Tom-Elliott : After executing boot.php. Let me go take a picture of it with my phone.

-Dustin

Tom Elliott

You might need to try:

http://fogserverIP/fog/service/ipxe/boot.php?mac=<macofhosttryingtoboot> (Of course replacing the <macofhosttryingtoboot> with the mac of the host trying to boot)

dholtz-docbox

Yeah, that was what the output below was from, actually.

#!ipxe
set fog-ip 10.1.10.42
set fog-webroot fog
set boot-url http://${fog-ip}/${fog-webroot}
exit

edit: I hit ‘s’ to enter PXE before booting, otherwise it stated that chainloading failed…

-Dustin

Tom Elliott

@dholtz-docbox Can you check the apache error logs then? If the error is still “chainloading…s to continue” or whatever it is, likely there’s some error being displayed that’s “breaking” things right now.

dholtz-docbox

@Tom-Elliott : Okay. Let me purge this log and get a clean log for you.

-Dustin

dholtz-docbox

@Tom-Elliott : I unfortunately have no errors in my /var/log/apache2/error.log file after cleaning it and running through the process again. Would they be generated somewhere else?

-Dustin

dholtz-docbox

I tried playing around with other types, and none appear to work. The machine uses GRUB boot loader, so I thought I would try to boot into GRUB, but it leaves me in the GRUB interface where I appear to be able to do nothing. I am not quite sure how to configure this so that it boots into the drive directly when no tasks are present.

-Dustin

dholtz-docbox

I have new errors after reformatting the machine in question and starting over. To iterate, the machine has two drives, is running Ubuntu Server 14.04, and is a LVM installation. The first drive is ignored, so I am using Single Disk - Resizable, setting its Primary Disk to /dev/sdb. Last, the host is set to exit its BIOS w/ type SANBOOT.

If there is a task to capture an image, there is no issue from what I can tell. The machine boots into Partimage and begins cloning /dev/sdb without issues. However, when I go to boot into the system w/o tasks, the system NEVER boots properly.

So… the new error. Upon reformatting, I switched the BIOS exit type back to SANDISK - after exhausting the initial selections prior. This time when I boot into the system, I am prompted with the following messages…

Booting from SAN device 0x80
Boot from SAN device 0x80 failed: Exec format error (http://ipxe.org/2e852001)
Could not boot: Exec format error (http://ipxe.org/2e852001)
Could not boot: Exec format error (http://ipxe.org/2e852001)

Looking into what this error means, so far I am reading that it regards the kernel. This doesn’t make sense to me. I am not sure why the system can’t just resume its boot process, I guess; or what the impediment is - when using BIOS exit type = ‘EXIT’ - if that is the standard course of action.

-Dustin

System Hangs - Boot from SAN device

Server

Description

183

11.6k

17.1k

154.6k