I totally forgot about it being in Virtualbox and that it had a virtualbox name for the disk id. That would definitely cause issues when you try and boot it normally since that id would only be present in a virtualbox environment. I'm glad you figured it out, because I don't think I would've been able to help further.

And yes, the live usb installer *should* be able to install to any target on the system. If it doesn't I would take up the issue on Eric's blog, as he tends to prefer that medium to provide support for his software.

1 members found this post helpful.

So, after some days after accomplishing the task to have en external usb Slackware hdd, and an "accident" in the mean time, summarizing the process I believe there are 3 key points one should pay attention to.
First, one has to have inline all boot definitions and use UUIDs instead of /dev/sdxY so to have absolute names of reference (fstab, initrd, grub/lilo, etc.)
Then, at least for my case, the scenario works with the huge (plus initrd) and not with the general kernel.
And lastly but of equal importance, one has to specify the correct modules in initrd so that the system will be able to recognize usb and load from it.
At this point I had a frightening experience when a slackpkg install-new followed by a careless slackpkg upgrade-all upgraded the kernel from 4.4.14 to 4.4.38! So I needed to make a new initrd but I forgot the modules I used previously in the successful attempts. And all the nightmares of the previous failed efforts came back to life...
Anyway, finally I found that the following modules do the job. I don't know which one -or more- of them is the savior, but now it works OK as with the previous kernel. And not only that, but I was able also to boot from this disk using another laptop -usb2- (whereas in the previous case I was able to boot only from the laptop I had used though out the whole process -usb3-).
Again, thanx to all who helped me to accomplish this task.

@bassmadrigal
At some point, when you showed me how to make an initrd from another linux distro (with the mountings etc.), you have mentioned some bind mounts. Previously I was attempting to make the initrd from another distro without these bind mounts (I did only the mount /dev/sdb1 /mnt) and it failed. Why do we need these? Aren't these directories exist also in the /dev/sdxY that we mount to /mnt? Why do we need to bind mount the other's linux distro directories to the /mnt also?


Regards, Apostolos.

1 members found this post helpful.

This is one reason I usually recommend blacklisting the kernels from slackpkg. Even Pat recommends blacklisting them (even though they're enabled by default).

My normal recommendation is to installpkg (not upgradepkg) the new kernels, then you can go through the process of building your new initrds and updating your bootloader. Then you can try and boot the new kernel. If it doesn't work, you still have your old kernel to rely on until you can get the new one fixed. Once you've verified the new one works fine, you can removepkg the old kernels.

Keep in mind, the default output for the mkinitrd command is /boot/initrd.gz, so I'd recommend changing that to match your kernel, something like /boot/initrd-4.4.38.gz. This would prevent you from accidentally overwriting the old, working one when you build the new initrd.

These directories contain information about the running system, this includes many files that cover kernel parameters as well as all your device nodes. If these aren't available, many applications won't function properly, especially if they're trying to get a running snapshot of the system (like mkinitrd would need information from the running kernel to be able to build an initrd).

2 members found this post helpful.

I have had this problem of being unable to mount a root from an external USB (hard disk as well as pen drive) for quite a while. As one can see in https://www.linuxquestions.org/quest...ck-4175689683/

It has come up again now when I tried booting Slackware64-15.0 from a USB external HD. I have spent 4 very long days (which included MUCH of the nights ) searching and reading.

This OLD thread is by far the best material I have found. I still have to read it a couple more times to be able to digest it. Things are presented in such a clear, understandable way. Even if it does not solve the problem, which I really feel it will, it has already offered a rare understanding of the processes I am trying too deal with. Really great. Thanks for that !

Have you tried "-w 10", as described here?

@ I did, but I don´t remember in which context, there have been just too many combinations.

Thanks for reminding me, I will sure try it again now after having read a good deal more.