Post Download Scripts - Not Executing
-
First I can say that post install scripts do work.
Lee also has some excellent examples of post download scripts (again for windows targets) but you can also get a good idea what could be done: https://forums.fogproject.org/topic/4278/utilizing-postscripts-rename-joindomain-drivers-snapins
I might put some output in your script so that you can see if it is executing on the target computer. Even a simple echo we are doing this or that.
I can tell you some things that will help you debug post install scripts.
-
Use a debug deploy task. Then pxe boot the target computer, after the target boots it will drop you to a command prompt. You can key in
fog
to start single stepping through image deployment. When you get to where your script should run hit ctrl-C to exit the fog script. Then you can take a look at your environment and debug your code. -
When you are at the command prompt of a debug deploy. If you give root a password with
passwd
you can connect to the FOS Engine via ssh (so you can copy and paste text via putty or what ever ssh tool you use). This help with debugging because you can paste in your script step by step to ensure that everything is working. -
Remember that your script is running from the perspective of the target computer using the FOS kernel so if you need something that is on the fog server, you will need to connect to the fog server via nfs and download it.
-
The post install scripts runs as root in the FOS engine OS. That has nothing to do with the fog user or the root user on your FOG server. The FOS engine is a complete stand alone linux OS.
-
-
Yes, that is the other article I am referencing - a lot of examples in it! Let me try adding in some echoes and attempt a debug deploy. I have never done a debug deploy, do I need to enable the PXE menu to get this to work? Currently I have the menus hidden. Giving it a whack in the mean time.
Thanks!
edit:
A question regarding bullet point #3 and #4 in your comment.Would this include copying the actual script to the host machine? Or, would this only include additional resources I import during the scripts execution - similarly to the fog/funcs.sh in the other scripts?
-Dustin
-
@dholtz-docbox No menu needed for debug deploy. Just schedule a normal deploy, but in the task create box selected the checkbox for debug. Then the FOG server will schedule the task. PXE boot the target computer and it will do the rest. You should see a couple screens of commands where you have to press enter, but eventually it will drop you to a command prompt. You need bits of the fog script to run to properly setup the target’s environment but you can break out after the image has been placed on the target “your echo command may be the clue when to exit”.
-
Debug output, like @george1421 mentioned, should help you diagnose the problem. Debug deploys can help you get it, but you can also cheat it by waiting for input:
#!/bin/sh set -x . ${postdownpath}update_swap_uuid.sh read -n1 -r -p "[$(date)] press any key to continue"
Off hand, I’m guessing it’s probably that “${postdownpath}” doesn’t end in a trailing slash, so your client is trying to execute something like /images/postdownloadscriptsupdate_swap_uuid.sh which likely doesn’t exist.
-
@mrayzies : I like that update to the fog.postdownload file, prompting input! I was avoiding a lot of these prompts, not sure how the script was executed and whether it had to avoid blocking commands or not.
Regarding the ${postdownpath}, I have tried with both a trailing slash and not, as I have no idea where this variable comes from. That said, where can I determine this variable’s assignment?
-
@dholtz-docbox When you break out of the fog deploy script. You can check out the entire environment. Just use
set | more
orset | grep postdownload
-
@george1421 : That makes sense. I am going through the debug deploy task as we speak - very cool! As soon as I can, I will query this (currently in the middle of Partclone).
I really want to learn more of what you guys did here, it’s really awesome. As someone overcoming the whole understanding-networking-in-depth, going through the process of learning what you guys have done has been great. That said, boy am I ready to develop something where the network doesn’t kick my ass left-and-right!
-Dustin
-
Okay! My script is being called, and I believe I see the issue…
Every time I deploy a new image, my swap drive has the incorrect UUID in its fstab. So I thought I would just update it after imaging, which led me to post download scripts. Now that I am looking at it, it appears that the ‘fstab’ has no contents at the time of its execution, so it is not able to update the swap drive’s UUID in the fstab after imaging the host machine.
Everything said, there is no way what I am doing is valid. I have seen very little resources on people having this issue, yet it is the very first thing I have hit with every image I captured. What can I do to ensure the swap drive gets assigned the correct UUID? When going through the debug process, I noted sections where the following occurred…
- Resetting UUIDs…
- Resetting swap systems…
- Restoring swap partition
For some reason, I feel like there is an issue in how I am capturing images, or in how the server resets the swap drives. I have to imagine it’s on my end though, given the nature of the issue and lack of articles with other users having the same issue. Is there a better way to handle how the swap drive gets assigned its UUID in it the host machine?
-Dustin
-
@dholtz-docbox This is really a question for the @Developers.
But what I wonder is why (if the UUID’s being wrong is the issue) do you not use the physical partitions reference and do away with the uuid reference all together. ?? But this may be just a simplistic thought to a complex question. The Developers would have more insight on what FOG does when it deploys a unix defined target.
But I can say the skills you just learned will help you when you want to do more advanced things like rename a host or change service boot states via the post install script. So all is not lost.
-
The following might be lies, so take it with a grain of salt til someone more qualified comments:
When your scripts are running, they are running from the perspective of the client; however, the file system you see is not the actual file system on the disk of the client; it’s a temporary file system that comes from the PXE file/kernel you boot. So when you operate on “/etc/fstab”, you are making a change to the fstab of something that is running memory, and thus no changes persist. If you really want to operate on files on the host, you need to mount the disk device. This is not an entirely fool-proof way, but for my scripts, I mount /dev/sda to /mnt and then can operate on files within that and those changes end up in the host. So for your example, you’d need to do something like this in your script:
#!/bin/bash mount -t ext4 /dev/sda1 /mnt ORIGINAL=$(grep -v '#' /mnt/etc/fstab | grep swap | cut -d ' ' -f1 | cut -d '=' -f2) ....
/dev/sda has worked well enough for me, though I am sure there are situations in which it would fail.
Secondly, to the heart of your issue, I believe your analysis is correct in that the UUID issue isn’t something you should be fixing in postdownload scripts. UUID’s, if I understand correctly, are unique, based on the hardware ID of the device. To my understanding, FOG should be handing that during the imaging process (similarly to how it handles growing the file system).
-
@george1421 : It could very well be a simple answer to an otherwise superfluously complex problem - of which I may have fabricated :P. Let me look into other ways of handling the definition of the partitions in the fstab.
I absolutely feel the problem is significantly simpler than I am looking at it from, so this would be a very nice solution if it works. I have no attachment to the UUID’s, I just need all the partitions properly identified, so that Partclone can capture the image and so the drive is registered when the machine boots. In the former, it appears to completely thwart Partclone if any of the drives appear to be “not mounted”. The later causes the machine to hang during boot, prompting the user to either skip drive mounting or to wait for the machine to figure it out.
Nonetheless, I agree. I believe the post download scripts knowledge will come in great use here soon We have a lot of plans for remote imaging, and I am looking to become pretty intimate with the everything FOG provides.
-Dustin
-
I guess I can jump in then?
The swap UUID is already set to what was originally taken. I’m fairly sure this is the expected case. But, for your /etc/fstab file I would recommend removing the UUID for the swap and define it for the partition itself so as to avoid confusion.
-
To add on to what @mrayzies stated, you do need to mount the individual partitions and use the mount point to access the etc/fstab.
As noted in the
ORIGINAL=$(grep -v '#' /mnt/etc/fstab | grep swap | cut -d ' ' -f1 | cut -d '=' -f2)
-
Sorry to spam, but I do need to say it.
I think your postdownload script was operating properly, though I’ve not tested it myself.
You are trying to edit /etc/fstab and as @mrayzies stated, this is the in memory filesystem, not the filesystem of which you’re trying to fix.
-
@Tom-Elliott : That makes sense, regarding the mount. This was also true for the values returned by my CURRENT assignment, via. blkid, where its UUID also did not reflect the host machine’s swap UUID. What I am interested in now is what you mentioned about removing the UUID for the swap and defining it for the partition itself. Would you mind elaborating on this a little for me? I am digging around trying to understand more about how to configure this, and I feel you are hinting on something similar to where I should be headed.
The image I am handling by-the-way, currently, is just an Ubuntu 14.04 desktop installation. So I have done nothing special to it other than install it, shut the machine down, and capture its image. I figured this would be the best starting place for a lot of things, but it is starting to sound like that is a loaded gun. That I need to configure a few things at a minimum to ensure subsequent images are deployed more smoothly?
-Dustin
-
@dholtz-docbox I need to see the fstab in question, but typically the UUID is used as the “reference” to the device/partition what have you.
So instead of setting fstab for swap as:
UUID=SOME-UUID-HERE swap swap defaults 0 0
You could set as:
/dev/sda2 swap swap defaults 0 0
-
Each of our machines look similar to…
# / was on /dev/nvme0n1p2 during installation UUID=61d640da-5df9-4a71-b6bc-cc28d8a8c9c8 / ext4 errors=remount-ro 0 1 # /boot/efi was on /dev/nvme0n1p1 during installation UUID=030B-0954 /boot/efi vfat defaults 0 1 # swap was on /dev/nvme0n1p3 during installation UUID=f0207d3c-a9b2-492e-93ca-fe37a59473d6 none swap sw 0 0
Are you saying that I can just change these to…
# / was on /dev/nvme0n1p2 during installation /dev/nvme0n1p2 / ext4 errors=remount-ro 0 1 # /boot/efi was on /dev/nvme0n1p1 during installation /dev/nvme0n1p1 /boot/efi vfat defaults 0 1 # swap was on /dev/nvme0n1p3 during installation /dev/nvme0n1p3 none swap sw 0 0
… and be good-to-go? Then image with the drives labeled as such, instead? I have been looking around for more details on doing something like this, but all I can find are people who lost their drives UUID and need to re-assign it. It’s overwhelming the amount of issues people have in simply manually assigning it, it really drowns out any other questions about fstab and what you can do with it and when.
-Dustin
-
@dholtz-docbox Yes, you should be good to go. Now I don’t know which partition is which.
One of the things FOG does automatically though, is reset the UUID to match of what the OS was expecting. So I don’t know why you need to change the UUID to begin with.
-
For this image, what do you have defined for “Image Type” and “Partition”? Perhaps you have something misconfigured here which is why the UUID of the SWAP space is not set properly?
I’d guess that you should have “Multiple Partition Image - Single Disk (Not Resizable) - (2)” and “Everything - (1)” respectively.
-
Oh. My. Gosh. I have been using Single Disk - Resizable this whole time! I thought this was the correct configuration, one DRIVE, which is resizable for each partition. When did this change, out of curiosity? When I first downloaded the FOG Project, I swear the multiple partition selection did not exist. This has me wicked excited! I was just about to go home too. Time for one more test!
-Dustin