[crux-devel] ANN: CRUX 2.5-test2
clemens.ml at gmx.net
Tue Oct 28 19:46:57 UTC 2008
>> CRUX-2.5-test2 seems to work here... even after doing strange
>> thiongs like this:
>> I am looking for a way to install CRUX from an already
>> booted foreign linux system without the necessity to boot
>> from the CRUX CD.
>> Sometimes I cannot access the console (hosted machine), there
>> is no CDROM-drive (netbook) and no quick way to boot from
>> a CRUX USB without building one manually.
>> Currently, I mount the .iso, unsquashfs crux.squashfs, copy /cdrom/crux
>> to the unpacked squashfs, chroot into it and run setup from there.
>> Well, everything seems to work fine... the only thing which I need
>> to install manually is unsquashfs from squashfs-tools.
>> 1) Does it make sense to include unsquashfs with the CRUX CD?
> Hmm, not sure what you mean. A squashfs package on the iso or the
> unsquashfs as part of the binaries available after booting the iso?
I think I need unsquashfs as part of the binaries available after booting
the iso. It would be also possible to unpack the squashfs-tools package
when it's included on the CD but that would need an additional step.
> In both cases it dosn't solve your problem IMO, because you either
> need unsuashfs to access the iso, or a running crux system to install
> the package. Or do I miss something?
Goal: No need to boot from CDROM, no need to access the console,
Assumption: A system always has a spare partition for a version of CRUX.
Usually, two small partitions w/ <10G each for / are sufficient for
That's roughly what I did (out of my head) to minimize downtime
of the running system:
- download iso.
$ mount -o loop crux-2.5.iso /mnt/iso
$ unsquashfs /mnt/iso/crux.squashfs
$ mkdir squashfs/cdrom
$ cp -a /mnt/iso/crux squashfs/cdrom
- chroot into squashfs
- mkfs.ext3 to some spare partition
- install to some spare partition
- configure crux and compile kernel (or copy from existing system)
- configure bootloader (usually lilo or grub)
- install custom stuff from existing system.
- cross fingers and reboot :-)
If everything works fine, a full system update is done with the downtime of one reboot. :-)
If something goes wrong, a reboot to the old system is possible.
It's possible to configure lilo as well as grub to do da failsafe-fallback
in case one system doesn't work.
Using kexec could shorten the downtime even more... well that's
a task for the advanced lection. ;-)
> Good question, but don't think so. Would you mind writing a short
> guide in our wiki, please?
Yes, once there is a working solution for the common CRUX user.
More information about the crux-devel