[pve-devel] Native ZFS and pve-kernel-3.10.0-7 question

Yannis Milios yannis.milios at gmail.com
Sat Feb 14 09:52:02 CET 2015


Hello all,

I am evaluating pve-kernel-3.10.0-7 on a standalone server with Debian
Wheezy+PVE repos(no subscription).

I have a local ZFS pool in which I store my VMs.The O/S is on a separate
lvm+ext4 formatted SSD disk(so no direct boot from ZFS pool).

Before using pve-kernel-3.10.0-7, I had to recompile each time spl-dkms +
zfs-dkms against new kernel (before rebooting) in order the system to be
able to "see" ZFS pool.

However this time I got a message that spl + zfs modules versions and those
located on the new kernel were the same so no recompile needed (type
--force to override).I decided not to recompile so I booted directly to the
new kernel which mounted the ZFS pool correctly!

My questions are:

1. From now on, we do not need to recompile anymore spl-dkms+zfs-dkms
against new kernels? (3.10.0-xx)

2. What should I do with ZoL repository which is already configured in this
machine?Should I remove  ZoL completely to avoid conflicts between their
version and your built-in kernel version? Can they safely co-exist?

3. As I understand the new grub will support directly boot from ZFS pool,
correct?
    Would that be a safe choice or maybe it should be better to have o/s
installed on
    separate disk(s) in BTRFS format which is natively supported by linux
kernel?
    I know it is not that stable but for just holding the o/s (not data!)
it should be safe         enough.In any case the ability to rollback o/s in
case of update/human error it               would be a nice option! :)

Regards,
Yannis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.proxmox.com/pipermail/pve-devel/attachments/20150214/3683b070/attachment.htm>


More information about the pve-devel mailing list