Chainloading failure: Toshiba Tecra C40-C UEFI & Samsung SSD MZNLF128HCHP-000

[george1421]

@mabarton Just so you understand FOG is a complex orchestration of a number of other open source projects. All must work together to produce an effective deployment.

FOG uses PXE, TFTP, FTP, PARTCLONE and NFS to be able to image files. The FOS engine (the software that is downloaded to the target computer to capture and deploy images) moves files over (unix) NFS connections (akin to a MS Windows file share) to the /images/dev directory on the FOG server. Once the upload is complete and the image information has been updated in the FOG database, the files are moved from the upload directory in /images/dev/<mac_address> to the /images/<image_name> directory. The FOS engine could do this over NFS, but that means reading the entire image back to the FOS engine’s memory and then writing it to the proper directory on the FOG server. But instead of doing this, the FOG developers are brilliantly using the FTP service on the FOG server to move the files about directly on the FOG server. To do this, once the files are captured to the FOG server, the FOS engine connects to the FOG server over the FTP protocol to direct the file moves. (This is the part you ran into yesterday). The FOS engine must log into the FTP service using the fog (user) maintenance account. In your case the fog (user) home directly did not exist, so linux block the fog (user) login via the FTP service. So the files were on the FOG server, they just couldn’t get to the final destination or recored in the database that the upload was complete. Once you created the fog (user) home directory the FOS engine should have been able to move the file from /images/dev to /images/<image_name>.

I know that was a long (and some what tedious explanation) but I hope you under a little bit more about the sequence of event.

[Wayne Workman]

@george1421 iPXE, Apache, in some cases ISC-DHCP or dnsmasq, MySQL, MariaDB, PHP, PartImage, probably others I’m not thinking about.

[Wayne Workman]

httpd php php-cli php-common php-gd mariadb mariadb-server tftp-server nfs-utils vsftpd net-tools wget xinetd tar gzip make m4 gcc gcc-c++ lftp curl php-mcrypt php-mbstring mod_ssl php-fpm php-process

And, others that are not listed which are directly downloaded and not installed through the repo manager.

[mabarton]

You guys make my head explode. Lol

It amazes me the amount of knowledge both of you have about Fog. Someday, I hope to have a fraction of that knowledge.

I will be at the school in about 2 hours and will check if the image uploaded.

Another question…does it matter where you install Fog and its subsequent updates? For example, do you have to be in /root or can you be in / or another location?

I ask this because I see reference to
/root/ fogproject, and I either do not have access to /root or nothing exists in the directory. Also, I am concerned that I messed up the install and that is the reason I am having this issue.

[Wayne Workman]

@mabarton It doesn’t really matter where the installation files are kept, but I recommend /root/git/fogproject or /root/svn/trunk generally. When running the installer you MUST have root properly sourced - meaning you need to BE root. In most cases, sudo -i will source root properly.

[mabarton]

@Wayne-Workman I use sudo su and sudo -i, but I have never used /root because when I use ls I see nothing in /root.

[Wayne Workman]

@mabarton Well, that just means it’s empty.

[mabarton]

@mabarton So, I am looking at the image of a 128gb SSD and it shows that the image is 460 mib. I am going to deploy this image to see what happens.

My concern is that the chainloading error means that Fog is not able to boot the SSD and so it is imaging something else???

[Tom Elliott]

@mabarton booting and imaging are two completely separate things. The size is more a convenience thing to let you know how big of a hard disk you’ll need. I suspect, however, simply disabling network boot after imaging is complete will allow the system to boot properly.

[mabarton]

@Tom-Elliott
That is excellent news! All of the other computers I have imaged with Fog have been legacy and HDD, and all of their images were almost the same size as the drive. That was why I was so concerned with the size/chainloading error.

Thank you, Tom. So, the chainloading error is more of a nuisance than a problem.?

[mabarton]

@mabarton So, is there a way that I can force the computers to shut down after the imaging process is complete? I am going to deploy this image to 100 computer and I don’t want them to boot loop until I get back to work.

[Tom Elliott]

@mabarton the chainloading IS a problem but I don’t have a way to fix it for now, so it is also a nuisance as it means, at least for this systems having issues, they won’t automatically perform tasks you might need. I think, however, this would be the perfect time to learn how to configure rEFInd to enable booting to actually happen without a problem.

[Tom Elliott]

@mabarton when setting up the task there is a place it asks you to shutdown after imaging.

[mabarton]

@Tom-Elliott Our computers go out to students living anywhere in Kansas, so I don’t get to use all of the benefits of Fog. Since the computers are not in my area, I always shut off PXE boot. Our brick-and-mortar uses iPads, primarily.

However, our school belongs to a network of schools called “Teen” and I am hoping to introduce Fog to those schools and a few others in our area. This is Kansas after all, and they don’t like giving schools money.

@Tom-Elliott said in Chainloading failure: Toshiba Tecra C40-C UEFI & Samsung SSD MZNLF128HCHP-000:

@mabarton I think, however, this would be the perfect time to learn how to configure rEFInd to enable booting to actually happen without a problem.

I am googling that now.

[Tom Elliott]

@mabarton don’t lack on sharing what you find. That way, maybe, I could provide a usable Config for all. After all open source is all about contributing. It sounds selfish I suppose but I would, and possibly many others, be most appreciative of yours, or anybody else’s, findings on this.

[mabarton]

@Tom-Elliott I fear that I would not be much help, but I rarely give up!

The deployment worked!!! I think I am going to have a party to celebrate the end of an entire week of battling this problem. I may be able to leave the school at a decent hour next week!! If it were not for this forum and all of your help, I don’t know what I would have done.

Thank you, thank you, thank you, Tom, Wayne, and George. Your help and expertise have been immensely appreciated!

[george1421]

@mabarton Sorry I’ve been out enjoying the holiday. So image deployment worked, you are able to clone your image to 2 or more target systems right?

The next bit you need to work on is the pxe exit mode stuff (if you want to). Just to recap you’ve selected all of the exit modes for UEFI Exit and none worked. Building on what Tom posted rEFInd should work. While I don’t have personal experience with this issue (yet), if rEFInd does not work out of the box (should be near the end of the list of options). There is a configuration file on the fog server that may need to be tweaked (refind.conf). Its located in /var/www/html/fog/service/ipxe on the fog server. We may need to add additional resources to look for (using the scanfor tag).

[mabarton]

@george1421 it has deployed beautifully-- with the exception of the boot looping:)

I am working in the rEFInd, so we shall see if it works.

Thank you for the support!

[george1421]

@mabarton If the default out of the box refind.conf does not work look for a line like this one ( of course the next time you work, and not on your free time).

scanfor internal

This tell refind where to look for bootable devices. You may want to append/change that line to search in a few more places, like this one.

scanfor internal,hdbios

IF that doesn’t work then extend the search a bit more (reminder you need to reboot the target after each change.

scanfor internal,external,hdbios,biosexternal

One of those should pick up that M.2 disk. (said with some hope) We are walking on some uncharted ground here so there will be many trial and error tests until we get it right (or just give up )

ref: http://www.rodsbooks.com/efi-bootloaders/refind.html

[Sebastian Roth]

@george1421, @Wayne-Workman Sorry, I’m just being a bit lazy after having been offline (visiting a friend in the mountains) for almost a week I am trying to catch up… Is this solved or are we still seeing an issue with this laptops?