New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Ok, I need to upgrade from wheezy. This is on a Hetzner dedi with 1tb root partition and 2tb user partition (2x3tb raid-1), both partitions pretty full. User files on root partition have a good backup, contents of user partition are backed up in scattered chunks of cheap storage that would be something of a nuisance to restore from.
I think of doing a reinstall to Buster instead of a multistep upgrade wheezy->jessie->stretch->buster which is likely to go wrong someplace. Can I do that without wiping out the user files on the root partition?
Alternatively I might like to reorganize the partitions so the root partition is say 50gb, then two user partitions (950gb and 2tb). There was/is some restriction that stops a single partition from being larger than 2tb.
Is that best done by just clobbering the existing root partition and restoring it from backup? That's a slow operation because of having to move a tb of data over the internet but it's basically a big scp/rsync that hopefully won't need much attention. I'll md5sum and check all the files at both ends before doing that.
Finally, one set of backups is on a Kimsufi with 2x2tb drives in raid-1 config. Since it is backups (and there's another set elsewhere) I might split that into two non-raid drives with 4tb total space. Can I do that without wiping the data? I might even want to do that with the hetzner box, since it is mostly static data.
Thanks for advice.
Get your shit together @willie ...!
EDIT2:
... one step at a time
patient and methodical-like
I think that multi-step upgrade may be worth a try, tedious as it may be - can always do the big bang reinstall if necessary afterwards
in any case: plan each step, take notes
and always remember - "when you're up to your ass in alligators, it's easy to forget that your original intention was to drain the swamp"
but I'm sure you knew all this - just some friendly reminders
DEBIAN SELEKTER OK !
EDIT3:
depending on what your time is worth to you, may be worth it to pick up a "scratch" server from the Hetzner auction to do a clean-slate install and consolidate your backups. Then move all that back over to the server you want to keep for sentimental reasons or whatever.
Yeah I think a lot of wheezy and maybe jessie repos are offline by now, making big bang reinstall look better. Plus I like the repartitioning idea, since fear of incremental upgrade failure messing up my root partition has been one of the issues stopping me from upgrading earlier.
I do want to consolidate my backups at some point: that is/was part of the plan with the new-ish 2x2tb kimsufi.
I like this Hetzner server because it had 2 new disks when I got it, that have held up nicely though they're no longer new. Getting another auction server likely means worse disks.
Yes I might pay the 10 euro for a month of 2TB Hetzner StorageBox to hold backups while doing this madness.
What is debian selekter? I saw selektor which is tor-related but I don't need that.
Thanks
So, no last minute surprise, the mini.iso (for us anal-retentive minimalistic pedantic purists™) hasn't changed since their last update.
http://ftp.debian.org/debian/dists/buster/main/installer-amd64/current/images/netboot/
edit: beware of multi-step/indirect upgrades, I've seen (minor, configuration-related) issues surfacing on debian-bugs e.g. apparmor profiles blocking what used to work. Or stuff disliking the new UsrMerge (links-based) hierarchy...
DEBIAN SELEKTER is a fresh new LET meme my man! That is all.
Thinking a "scratch" auction server would be just for a month (unless it turns out you like the new one better - ala the Monty Hall problem) - I guess something more like €25 may be not such a quick decision to make just for personal stuff though.
Anyway, whatever you do - have fun.
Yeah I just don't see how the scratch server would be better than a normal backup (to storagebox, cheaper per month than another server), followed by full reinstall and restore on the original server.
Thanks. Debian.
@willie indeed ... I do tend to talk myself into buying a bit more than necessary from time to time lol.
this is how the internal debate might go:
D. "well gosh, what if the Buster setup takes longer than I expected. I sure would like to have my existing setup available in the meantime."
A. "aw, c'mon - don't spend extra money - try the incremental upgrade. Everything will be fine."
D. "yeah right. The devil is in the details. Let's at least make sure we can restore everything from backup. I can haz storagebox?"
A. "mmmm €10 I guess that would be a prudent expenditure all things considered."
D. "alrighty then. If €10 is acceptable then why not another €15 hmmm ...? We can keep original online, consolidate backups, and have a fresh install ready to go all tested in case the incremental upgrade on the original server goes sideways somehow."
A. "no way - €25 just for a scratch server? That's gratuitous and excessive."
D. "but but but ... what price satisfaction? Maybe the new server will have even better disks - auction listing sez Enterprise HDD!"
A. points to some other shiny object to distract while hiding wallet
By the way, a few weeks ago I installed Buster (testing) from ISO when I noticed the option on one of BuyVM's new nodes in Luxembourg.
Also noticed @freerangecloud currently has an "alpha5" version of the Buster ISO available (setting up a new instance in Fremont - assume they may have it in other locations as well.) Hopefully they'll put the stable release ISO up soon too.
First thing that comes to mind is this:
-Check all backups
-Split the raid (just --fail one hdd and leave it degraded, wipe partitions/raid from "failed" HDD)
-Create new raid1 (degraded, with 1 "missing" device) on "failed" HDD, install new OS, copy data.
-Destroy original raid, add second disk to new raid, resync.
If everything goes well (both HDD-s survive i mean) it should be relatively fast, avoid transferring data over internet and allow new clean install with new partition structure.
Thanks Gamma17 that sounds like a very good approach. I will try something like it on the backup server first.
@uptime @willie Hetzner cloud : 3hour experiments for just 0.01.
Get it out of your system. 🤞
Yeah I was just thinking about that, wondering what hourly vps let me upload an iso and run it. I'm not sure Hetzner has that but I'll check.
Hey @Hetzner_OL you got Debian 10 loaded as a cloud installation choice yet ?
Hmm, I still have a dedi with proxmox. Maybe I can figure out how to use that.
Jessie is still supported with security updates, so the Jesse repo is still available. In the case of Wheezy, the repo has been archived, but you can still reach it via
deb http://archive.debian.org/debian wheezy main.In your case, since you also want to rethink the partitioning, I would probably do a backup of personal files followed by a repartioning plus clean reinstall.
I spun up a DO droplet, found that I can install custom images in .gz format etc, but not .iso, so launched with debian 9.7 (the only version offered) with idea of upgrading from overloaded debian repo. In fact DO's local mirror already has buster so the upgrade is going super fast. Seems to be progressing ok so far.
Added: upgrade seems to have worked, rebooted, playing with the droplet but will shut it off soon.
Lunanode, Hetzner, Vultr, and Linode all suppprt uploading own custom ISO
Also vmhaus but would need to put in a ticket (last time I checked at least)
EDIT2: or use netboot.xyz
It's like you can always hear a ticking clock on those pay by hour VPSs.
ISTR that custom isos with Hetzner require some kind of contortion involving the rescue system and a second Hetzner server (though I do have a couple). I'll take a look at it and maybe try Vultr, though there's not much point. Main thing I should care about is whether Hetzner and OVH have Buster on their local mirrors yet. But if they don't now, they will soon.
I did shut off the droplet after 38 minutes of runtime. DO now bills by the second so I saved a fraction of a cent compared with their hourly charge
. The droplet was very responsive compared to LET stuff I'm used to, fwiw.
Any VPS or dedi with a VNC/Console can install from ISO. Just use grub-imageboot. No need for weird upgrade paths. And if using the latest and greatest netinst/mini.iso, you'll have an up-to-date system right from the start.
In case of dedi it is always possible to just launch own kvm, attach hdd directly to it and install anything needed. Then shutdown VM, reboot the server from those hdd and that's it.
Helps with reducing downtime too, since new os can be installed/configured without shutting down old one first.
Well, upgraded my Virmach small kvm, seemed to work, rebooted to run the new kernel, and now it won't let me ssh in. It's late here and I'll mess with it tomorrow. There was no local mirror so the upgrade came from ftp.us.debian.org which as you can imagine was rather slow at times.
Added about 5 minutes later: ok it's up now, it was just pretty slow to come up compared to what I'm used to.
Upgraded from Debian 9 LXDE to Debian 10 LXDE. No big issue. So far I only found a small issue with Slack Desktop (v. 3.4.2), which displays some errors when clicking on links inside.
vim 8.1, Rustc 1.34 and Python 3.7.2 nice!
https://www.debian.org/News/2019/20190706
I've been using it for a month or so on some of my servers, plan on installing a fresh copy on my Worldstream dedi today though.
anyone got any news about them shutting down the jessie onces? we guess they will be slowly killed
Jessie is on LTS until 30 June 2020, so the Jessie repos will be available until then.
alright, well should wget -r it