I just switched from 32Bit to 64Bit Kerner using this Tutorial https://wiki.debian.org/CrossGrading.

However, I am missing the reboot command (probably a lot more commands are missing). And yes I am logged in as root.

How can this be fixed?

  • 1
    Be wary of following complex operations over "instructions" in the net without fully understanding them. Once when looked with attention, often a critical step or instruction is missing or it needs slight modifications for your specific case. I migrated my 32 bits servers to 64 a lot of moons ago with a similar procedure. Could you be more specific of what are the actual problems, and what it is up with the reboot command? Dec 8, 2015 at 13:33
  • @RuiFRibeiro Yes, I needed to this. There was no other option for me. I am missing a lot! of commands. I just figured that commands like rm, mkdir, ls and so are missing as well. Looks like the system should be reinstalled? BTW I did this with a fresh installation, no important data is lost.
    – Robin
    Dec 8, 2015 at 13:45
  • Could you show me the output of "uname -m", and the output of "file /bin/rm", or they do not work too? Dec 8, 2015 at 13:47
  • @RuiFRibeiro uname -m -bash: uname: command not found and the file /bin/rm is not available.
    – Robin
    Dec 8, 2015 at 13:49
  • 1
    "echo 1 > /proc/sys/kernel/sysrq || echo b > /proc/sysrq-trigger" can be used to reboot a server. Dec 8, 2015 at 14:00

1 Answer 1


A 32 to 64 bit hot upgrade procedure has to be performed with some care.

Essentially what happened is that you are in the middle of the migration procedure and have a mixed environment between libraries of one architecture and binaries of another, and then some commands could not be installed, and others were the library installed do not match anymore the binaries. (e.g. 64 bits binaries with 32 bits libraries for instance)

After booting the 64 bits kernel, and upgrading the dpkg/apt binaries to 64 bits, the migration itself from 32 bits commands to 64 has to be done almost in one go in order to have a stable environment, either with a script, or with the help of statically linked binaries (or a mix of both).When I performed the migration, besides the instructions I used a mix of busybox, mc and debootstrap to allow me to correct mistakes made without losing control of the environment.

Once you lost control of the environment, the alternative would be to finish the migration with a live CD were it an important server.

If you continue to test this, eventually you will find the right steps to do a complete hot migration for 32 bit to 64 bits. (I am making the assumption you are doing tests)

Since echo is a built-in bash command, depending on the way the kernel was built, you might be able to reboot (as root) the machine with:

echo 1 > /proc/sys/kernel/sysrq
echo b > /proc/sysrq-trigger

Be aware the system is in an unusable state, and won't startup again, except from another boot partition or boot medium.

  • Thanks. Probably I will just reinstall the system. I have no physical access to the server, therefore the live cd will probably not work. Update: I cannot connect to the server anymore. Probably ssh is broken as well.
    – Robin
    Dec 9, 2015 at 7:45
  • 1
    More probably the shell, ssh should still be running in memory. I am not surprised, I am afraid. Dec 9, 2015 at 9:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.