HI,
first we never include in our tests migration between more than one level. In other words, we tested migration:
- from 9 to 10
- from 8 to 9
- from 7 to 8
- from 6 to 7
That said, we also have a bug opened by Ian few years ago asking if we really need to keep all the migrate scripts for the db files lying around for a lot of time. SO there should be there some hope to allow you to do 6 to 10, but I would do a test first.
The global idea would be to backup your original server to a tgz.
you could do it in different ways:
1- simply mount the original disk using another linux and chroot to do the backup from console
2- follow
https://wiki.koozali.org/UpgradeDisk doing a chroot to the disk to issue the pre-backup before the other step
3- try the migrate helper script after doing a chroot
4- (this one would be an alternative to tgz)chroot to the disk to issue the pre-backup ; then signal-event pre-restore your new server. rsync files from the the mounted drive using the usual list of path to backup ; signal-event post-upgrade and reboot.
the global idea is really get a backup smeserver.tgz in the way you are the most at ease and using conventional way to restore.
If you are lucky enough you can do it only once 6 to 10, other wise you will need to do it 4 times.... at that point, doing once or 5 including a failure, it is worth trying the all in one first... and report
remember to put the custom-templates aside and restore them one by one and see what is breaking because it is not a if but rather what will break in term of probability for 4 versions apart....