Это видео недоступно.
Сожалеем об этом.
How I upgraded and reverted Proxmox from 6 to 7 to 6 with UEFI boot and GRUB
HTML-код
- Опубликовано: 5 авг 2024
- I wanted to upgrade my Proxmox VE (www.proxmox.com) Server from version 6 to version 7. But I did not want to do this without a Plan B, a fail back plan. For this I used GRUB and efibootmgr. I converted a small swap partition into a bootable Linux partition and pivoted the Version 6 to Version 7. When I noticed that things did not work as expected, I was able to revert to version 6 in less than a minute using GRUB, UEFI Bios and efibootmgr.
The blog version is here: www.onemarcfifty.com/blog/vid...
RUclips: / onemarcfifty
Twitter: / onemarcfifty
Discord: / discord
Github: github.com/onemarcfifty
Patreon: / onemarcfifty
Blog: www.onemarcfifty.com
0:00 What's in this video ?
0:38 System overview (before) and THE PLAN
1:53 Considerations when upgrading a server
3:22 The disk and Volume mayout
5:00 Turn a swap volume into a bootable Linux Volume
7:20 How does UEFI Boot work ?
8:55 Update EFI Boot settings
12:03 update GRUB
13:00 Quick Summary
15:00 Shrink the original Volume
16:07 Creating the new Volume
17:10 Upgrade Proxmox (finally!)
17:42 Reverting back to Version 6
19:04 Some Cleanup
Just a little warning ;-) This video is all Linux command line ;-)
I wrote a nano RTOS for Cortex-M embedded devices. Compared to that the Linux command line is like a toy for children 😎
It’s interesting that proxmox as an appliance-like solutions does not come with multiple boot environments and atomic updates out of the box. But good hint to keep this partitioning scheme in mind.
Hi Bernd, yes - you are right. It _should_ or at least _could_ have that built in. I have however had so many bad experiences with in-place software upgrades in the past that I have really become over-cautious with regards to this :-)
Awesome vid ! previously i've used a grub repair bootable usb for repairing partitions...and its been many years now since i needed to use chroot when installing Gentoo systems.
Thanks to this "refresher" video, i am now peeking behind the curtains and am seeing how the boot process actually works. Great content Marc, many thanks for this.
Hi Damien, many thanks for the feedback. Before everything else I was curious to see if I could really do everything just by using ssh and not needing to go downstairs to the basement ;-)
Best it tutotial on RUclips. Was not planing to Learn the “New” boot method today but it was on my bukket list but now i have 😂 thank you vers Munch!
Always looking forward to your videos; Thanks.
Thank you very much.
This is really cool, it mimicks the "dual boot" function of many embedded devices, that allows to do system upgrades on one partition and can failover to the other partition if the new firmware upgrade does not work or has issues
Hi Marco - you are right! I had not even thought about that aspect. Many thanks for your comment.
These dual boot devices like are a great platform for driver and software development. I once made a mistake resulting in kernel panik, and it booted from the other volume.
Nice, excelente explanation of a complex process 😎
Thank you ;-)
Cool stuff. Great explanation. Good job. You're very clever.
Many thanks Gabriele.
Good video Marc.
Thanks a lot ;-)
Awesome! Simply awesome!
Thanks mate.
My mind just exploded
Ooops - in a good way I hope ;-)
Hi Marc! Sorry for the newby question, but: I assume this would also work for 7 -> 8 version upgrade, right?
I'm wondering if `parted` would do the math to calculate the "correct" lv size and issue adjusted e2resize and lvreduce commands? I've seen mention of `lvreduce --resizefs` don't know if that would help. If these tools don't work, I would have shrunk the volume to 35GB and then the lv to 40GB and then resized ext4 again to the auto-detected size. Sometimes adding or removing an lvm extent can align the end of the filesystem, so there are fewer unused sectors.
Hi Mike, great comment again. It's true - I have been generously "wasting" space here.
Do you think it's possible to make it all a large cron task so we didn't end up with a backup being too outdated?
Hi Abdullah, you mean to schedule a backup of the system and have that handy for restore if things go wrong? Yes you could. However - you'd still have to resolve the open files issue and potentially either stop services or create snapshots.
Hi Mark why or how did you choose openWrt over openSense ?
Hi Richard - It's mainly related to the hardware I used at the time - OPNSense needs x86, OpenWrt runs on many more platforms.
I think this method is similar to how smart phones and tablets apply OS updates, interesting guide. Nice idea. Meanwhile i prefer btrfs over ext4 plus lvm as i often transplant, move or change proxmox installs (for fun only). Btrfs is meanwhile well supported by gparted and gnome-disk, so i can manipulate with a debian live disk. Of course this is not your usecase here.
Marc, it would be much faster to use dd or other software for cloning disks and partitions, instead of rsync. Are there any benefits?
I wrote my own disk cloning tool, which has the ability to offline disks or lock volumes. It is very fast.
A few years ago, I needed to remotely perform offline maintenance on a couple of Hyper-V servers in my university. The approach was very similar: install Windows on a VM, setup bcdedit to boot once from the .vhd, and boot from that. Then clone the host partition and the next reboot takes me back to the original host OS.
dd clones raw partition, rsync moves files. If you have a 100GB partition that has only 5GB of files, dd will move 100GB of data, rsync will move 5GB of data. If your partitions are mostly full, then the difference is not that big
@@marcogenovesi8570 Good point, however on filesystems with many small files, it's usually faster to clone the entire volume. For example a 100 GB volume clones in 7 minutes at 240 MB/s.
another reason for using rsync is that it allows you to "clone to a partition of different size" while dd does a 1:1 copy
btw, if you want to show realtime speed metrics and a completion bar in the console when you clone stuff with dd, you can use the tool pv instead (or in the middle of the dd command, with pipes, like dd /dev/sda1 | pv > /dev/sdb1).
dd isn't special for a straight 1:1 copy as most of the work is done by the "everythign is a file" stuff in /dev anyway.
For example you can "cat /dev/sda1 > /dev/sdb1 " too and it would work fine at similar speed as dd optimized for blocksize, at least in my experience
@@marcogenovesi8570 Thanks! I use my tool gfc file compare, which provides statistics such as time elapsed and average speed. Multiple threads are utilised for read and write to achieve maximum performance. Offline cloning is very important.
`wipefs` should be used to destroy superblocks when switching a partition's type.
Hi Mike - you are right. If the disk had previously been used in a raid or the like then that would have been necessary. I never had issues with simple partition / volume types though. Also keep in mind that it's not a partition but just a LVM volume. Not sure how things would have worked with an EFI partition or the like (w/r to partition flags). Great comment, many thanks !
Is there a way to do this with just zfs?
Hi Michael, call me old fashioned but I am really trying to stick to simple file systems such as ext(2/3/4). Haven't given ZFS much thought I admit. Even though the features are amazing I always try to keep in mind how I can fix it if I break it ;-)
@@OneMarcFifty thank you for your reply.
ZFS looks really intimidating at first. But it is really amazing, someone accurately described it as "alien technology".
I've been working with it for the last 3+ years.
I really like adding sanoid/syncoid to proxmox because it adds a super powerful way of autosnapshots but even better sending data to another zfs machine which enables off-site replication.
this must have a easier way to do it... 🤔
Yes - of course. This is not the easiest and smoothest way of upgrading Proxmox. I was just curious to see a couple of things. Could I do everything over ssh and could I replicate a Linux to another disk on a file per file basis using rsync ;-)
OpenWrt Video Request: Block a List of IPs. ...using IPSet and Traffic Rules
Ah - good one again - as I said in a different comment - inbound filters is definitely something on my list ;-)
@@OneMarcFifty Not inbound filtering. I want to filter outgoing
Oh interesting- what’s the use case behind?
@@OneMarcFifty To block malicious IPs, troubled or high abuse rate countries
Ok got it, thanks.