Some questions/thoughts about last week's "problem" stable update [2016-10-9]

Hi, I follow what I saw someone call the "super stable" update method: read the annoucement thread and wait for any dust to settle, let others thrash it out. The latest one [from last week 2016-10-09] has got me worried though.

I read the whole thread, returning for new posts. I see someone just a few hours ago still having problems. A few thoughts and questions.

About 95% of the meat of that thread made my eyes glaze over. I think I may be over my head here. Migrated from MS 2 years ago, no chance am I going back, may need to switch to non-rolling type Linux.

I like this forum and the idea of Manjaro/Arch/rolling release etc, but for instance I have some important transactions going on with more coming. I cannot research or stop for any fixes at this time.

I am running Kernel 4.6.7-1

Question: Do "problem" updates [like this last one] get changed/fixed? Is it now relatively safe to proceed with the update? I use xfce, don't have any encryption, don't do any gaming. I saw many "no problem" posts, in 18 months I haven't really had any problems either.

PhilM's blog post linked in thread: "Update recommendation for SystemD users". Showing my ignorance here, I thought "systemD" was integral to all Manjaro install, not something you choose. How do I know if I'm running it?

https://manjaro.org/2016/10/10/update-recommendation-for-systemd-users/

Repeated mention of "tty" procedures with various numbers after it. After some research I'm stumped again. Is tty something different than the normal terminal? [edit: what I call "normal" apparently has been the terminal emulator on the desktop, had forgot or not fully grasped that previously. Explanations in further posts below, thanks]

In this update announcement and previous I've seen mention of and links to some kind of "test" media you can run before performing an update. I read some but again, no idea what is being presented. This is not a complaint, I realize one has to dig in to be able to do rolling release. Just re-stating what I've seen elsewhere, instructions can be really daunting/cryptic to beginners. You can run a test before updating that will not damage anything?

Just airing some thoughts out, thanks

Simply have a backup and recovery strategy for any update. Everything else is taking risks. On any distro, even non-rolling, which occasionally get updates, too.
I'm using btrfs snapshots, there are similar features in zfs filesystem or LVM volumes.

But even things like Clonezilla can be used (actually providing much more security), would take several minutes on a good machine.
https://wiki.archlinux.org/index.php/Synchronization_and_backup_programs
https://wiki.archlinux.org/index.php/Disk_cloning

3 Likes

Yes, thank you eugen-b, that is indeed the best "short answer" I am woefully lacking in such procedures, really need to get on it [perhaps desparately so :slight_smile: ]

Regardless what os you are using (rolling/non-rolling Linux, and most especially ms), there is no guarantee that anything is 100% "safe". Nobody can guarantee 100%....not in any realistic way. I am convinced that Linux is "safer" than ms....for a lot of reasons. That being said however, if you are operating "mission-critical"....back-up your data regularly to be "safe".

Best regards.

Thanks conky57, I did have some questions in there as well, here's condensed version

Is "tty" the same as the terminal, or is it something different.

How do I know if I'm running systemD? I thought it was integral to all Manjaro intalls, apparently not.

There is some kind of "test" media you can run that is like a non-invasive dry run of the update on your system? Available for all updates?
thanks

I think you are getting in to deep Bad updates are very rare, I use Arch for 15 years same install for over 10 years, can't remember a bad update in the past 4/5 years that is more than I can say for Suse Fedora and Debian.Also Manjaro for 4+ years. If you keep to the stable version you can't go wrong.
Systemd is the default for Manjaro there is a option for openrc net install and a user does a XFCE4 openrc spin.
Some users went over the top in that discussion amounting to ranting for the sake of it. The simple fact is win10 has broken on more updates than all the linux distros put together and apple breaks as well. No distro can be 100% unbreakable on updates their are to many varables,
My personal feelings are this there is no thing as a broken system that can't be repaired if you have the knowledge if it does break their are experienced users that will help It may take a little time mind you as users have a life, But count me out on that i'm dyslexic and old so it take me time to do things but if I can anybody can.
You should either back up your data or save to another partition that is what I do so I only save temp to the home folder if the system got erased 1hr and it would be back up and running, My main reason for saving to external drives is just logic I have my system running 18 hrs a day my other drives are active but parked most of the time

3 Likes

Using ctrl-alt-F2 will put you into a terminal that is outside the xfce windows environment. That "tty" terminal will require a separate login. The reason it is recommended to use tty is that the windows manager may crash during this particular update.

2 Likes

A "braindead" method would be to

  • have a separate home partition,
  • run a more or less standard Manjaro install
  • (or have a list of installed packages, pacman -Qqen > packages.txt)
  • a current Manjaro install medium
    => reinstall each time an update breaks your system and wait for a fix.
  • use packages.txt to install what is missing:
    sudo pacman -S --needed $(< packages.txt)
    (--needed should install only those which aren't already installed)
  • If the installer doesn't offer to use existing home partition,you can mount your old home partition in fstab
1 Like

It is on all official manjaro installs ... if you didnt opt out on purpose Im sure youre running it. You can just use a package manager to search for systemd to see if it is installed.
tty is not the terminal on your desktop. Its a bit like dropping to DOS (no gui) on windows.
It's not that scary ... you will just be doing a terminal-style update in tty. Here ;

Ctrl+Alt+f2

Login with username/password

Edit - to be extra safe you can install systemd components first:
sudo pacman -Sy libsystemd systemd systemd-sysvcompat

sudo pacman -Syyu

Remove qtchooser? - Y

reboot -n

The problem is a bug found in a version systemd shipped - which might baulk during update/reinit resulting in a half-update/half-broken situation. You avoid this by being in a more basic environment that will not crash during the process. After this all is rather well. I am on testing and have since used normal update procedures with no hassle.

For having less fear when using TTY or booting to command line prompt (which goes like this: press e at Grub menu, add 3 behind the words quiet splash, Ctrl+X) install some CLI programs like pacli (package management), elinks (web browser), ranger or mc (file manager), nmtui (connect to WiFi), mhwd-tui (install kernels and drivers).

In 231-4 ?

My understanding was that it was caused by 231-1

1 Like

The transition (the first seconds) from 231-1 to either 231-3 or 231-4 was the problem.

OK, thanks, went there and saw it. I'd have to look it up, but how does one get out of that back into your normal system? I had to hold the power button down to exit, then re-booted.

use the command

reboot -n

to restart. Your normal desktop will boot again after restart.
You could use 'startx' to start desktop .. but after update you dont want to, just reboot.

Yes, but I wonder if the issue will go on with 231-5+. I have no reason to update 231-1 I now run. I installed all the updates by pacman -Syu (including 231-1 → 231-4) in vb with no problem, though.

OK, thanks, so the only way out it to restart then [which I did bluntly with the power button]

1 Like

The arch warning was if you update both 231.1 and the kernel at the same time you will get a problem update 231.1 1st reboot but then the kernel 4.8 which is buggy as hell tried it in void and it kernel panicked and then It was announced by the man its a bummer

This didn't happen during the update from 231-3 to 231-4.

I see; many thanks. Why to do this in tty, then?

Forum kindly sponsored by