Closed caseycs closed 5 years ago
Same problem still exists with 4.14.12-mainline-rev1
Turned out to be spl/kernel incompatibility. The only way I could make it work it 4.10.8-std-1
bootscript with Ubuntu 18.
A workaround is to KEXEC the Ubuntu Linux kernel. See https://blog.simos.info/how-to-run-the-stock-ubuntu-linux-kernel-on-scaleway-using-kexec-and-server-tags/
I managed to manually compile ZFS from upstream, but it was too fragile to trust it.
@caseycs Ilia, how did you manage to get it work with this bootscript? SPL module compile fails on my server. Thanks!
@michael-bouvy to be honest I gave up with zfs on scaleway few months ago, turned out to be too much hassle.
@caseycs thanks! Did you use LVM or other alternative?
@michael-bouvy initially I was using zfs for both lxd and docker containers, however I migrated completely to docker, so have no need to lxd anymore. Do not have any performance issues with native built-in docker fs driver.
@caseycs ok, I'll have to find another way to be able to do snapshots of my LXC volumes
It seems that the Start range doesn't have these kernel problems and a conventional ZFS install works fine...
@3ruce Scaleway have fixed their images and now they use the stock Ubuntu Linux kernel, which has built-in ZFS support. The issue is that you need to launch a new VPS.
Hi!
I'm trying to switch to new bootscripts, so I tried both
x86_64 mainline 4.14.11 rev1
andx86_64 4.10.8 std #1
However on each of them I could not install zfs kernel module:
Also I can't rollback to the previous bootscript, so currently ma machine got stuck.
I'm sure that same script worked perfectly before, maybe something is broken only for latest bootscripts?