I have had to patch deepin-daemon and startdde and have now forwarded them to testing and stable branches.
I experience no issues on my install. Update should be safe
Just make sure your mirror already has the very latest package versions:
[Stable Update] 2019-02-19 - Kernels, KDE, LibreOffice, Systemd, VirtualBox, Deepin, Qt, Firmwares, Wine
Everything went well except the Chinese translation is missing for ‘shutdown’ at the SDDM login screen and logout screen in Plasma.
Manjaro KDE with kernel 4.19 NVidia proprietary driver.
If updates from the Stable channel brings approximately 20% of users with problems why not create a new update channel more stable than the current one? I know Manjaro is rolling release but there should be a way to prevent users getting into those early problems from the updates. Maybe a channel called “Novice Stable”.
I’m running Deepin and upgraded, no issues found.
Updated on 2 machines via tty with no issues.
I would like to know how you think it would be like in more details? Honestly, I don’t know how it can be done without a massive amount of additional work.
Well i’m happy to be corrected if i have not remembered this correctly, but afair even if not in this update, there have been sad/angry/worried/confused posts for previous updates where users initiated their update in their graphical session, using terminal, when part-way through the update the graphical session froze or maybe restarted, leaving them in a bit of a mess. Again afair the risk is much higher for “critical” updates like
systemd, Plasma/GNOME/Cinnamon etc.
Consequently, once i learned of this hazard several months ago by paying attention to the posts of other Manjaroos much more competent than me, i instituted a personal policy [for my Unstable & Testing VMs, & my Stable VMs + Stable real Plasmas] whenever a new impending update involves those critical packages, of ceasing doing these updates any more with Octopi/Pamac, also eschewing doing them in Konsole, & instead always logging out of my graphical session, changing to tty2, then initiating the update there [via either
sudo pacman -Syyu or
pamac update depending on my mood & the prevailing wind direction]. I know that some cautious others go even further than me, but so far this feels like a comfortable & safe middle-ground.
FYI: Before Manjaro i ran openSUSE Tumbleweed, for which upgrades [as they call them, in lieu of updates] must be done in terminal or tty, & where noobs frequently broke their systems trying to do it graphically. Furthermore, even in something as vanilla as KDE Neon, i’ve had, overall, a really crappy history doing updates with the abominable Discover, & so i found that terminal or even tty proved vastly more reliable.
I’m happy to use Octopi or Pamac for all updates that “only” involve “app” / program packages [eg, Thunderbird, Firefox, Chromium, Vivaldi, Firejail, CherryTree, LibreOffice et al].
PS: Oh yeah, forgot to mention before… not applicable to any of my VMs, but for Tower & Lappy’s real Plasmas, i also ensure that the automated scheduled daily Timeshift snapshot did run before i will initiate the update. Also, i always do a data backup before every system update [ie, those involving
systemd, kernels, Plasma etc]. Yes, that’s me, a total scaredy-cat.
That remembers me Fedora when devs litterally recommended not to run “dnf upgrade” in a graphical session.
Still doing it graphically with Pamac GUI, even when it’s considered “dangerous”. You know, for science.
Truer words have never been spoken @kdemeoz, I learned my lesson the hard way awhile back. Tried to update via the GUI and borked my system… with NO backup. Now I have become a ardent supporter of and advocate for Timeshift. I also follow the same policy as you.
I know, i have found all your posts on this very interesting & laudable… i think you’re doing us all an excellent service with your tests. The thing is, i’m sure that you are infinitely more competent than me in rescuing a borked system, therefore knowing my limits [even though i have learned how to recover a couple of broken systems with chrooting (which btw made me feel very happy)], i made the philosophical decision to adopt the conservative strategy i mentioned. The old adage “better safe than sorry, when you know you’re a numpty”.
More than 400 packages later everything is fine on KDE. Update was done on tty, but plasma didn’t crash. Many thanks to the team.
My Bluetooth has stopped working. I don’t know if it is related to this warning that I got but
sudo chmod 755 /usr/share/polkit-1/rules.d/filesystem gives:
chmod: cannot access '/usr/share/polkit-1/rules.d/filesystem': No such file or directory
I do use Bluetooth to connect to some peripherals, so I truely appreciate any help to get it back working again.
Warning: blueman: directory permissions differ on /usr/share/polkit-1/rules.d/filesystem: 750 package: 755
My experience with Manjaro OpenBox installation
That’s interesting. Once i read @philm’s links re this alleged problem, i tested my LO Calc & Writer, vs Kate. I had no problems at all copying & pasting between Kate & LO, back & forth. I wonder what special circumstances need to exist for this alleged bug to actually manifest?
The thing is, as I said, I don’t know how I manage to survive on Testing by doing everything exclusively with Pamac GUI, both in VMs and bare metal. It just… worked.
I honestly though that it wouldn’t be that bad in Stable and that the systemd saga would finally end with an happy ending; I’m really surprised in fact that it turned out that bad.
Although I do not use Gnome or KDE (which seems to be the ones that suffered the most, although it might be biased because of their “flagship” status); I only use Cinnamon. (I had a VM in Gnome, but it was just a vanilla Manjaro Gnome with nothing customized, and it’s a VM too, not a unique PC with unique spec.)
We aren’t the only one that are having it rough. Many issues we are currently having are common to Arch Linux too. Here’s a detailed list (I gathered from Arch website and Reddit).
Problems with LibreOffice on KDE5
The duplicate global menu (that got fixed in Testing)
Crashes while upgrading (On Arch, it happened with systemd 239.300, but Arch never went back unlike us last december; they just went trough it as it was nothing. It seems that the transition between 239.6 and higher is just… troublesome. Perhaps the downgrade we did last december was an error?)
Systemd 240 was not necessarily a great moment for some Arch Linux users too (Like it wasn’t a good moment here in Unstable branch too)
Powertop failing with recent kernel
System on KDE Plasma eating a ton of RAM for apparently no good reason
TL;DR: Reading what’s happening on Arch forum, bugtracker and Reddit, thing are not really much better than here. We both have common issues and having a hard time.
The fact that Manjaro published everything in one big release make it looks like apocalypse happened yesterday. On Arch, since things get published pretty much continuously, problems come and go at different moments and at a faster rate.
Someone needs a hug
So if I update say tomorrow or I wait like 5 days to update, will the the packages still most likely be the same or should I wait a few more days for any extra last time package updates being pushed into stable?
Everything went smoothly! Thanks for all the hard work.
Just wait for next week and upgrade… Hopefully by then manjaro staff would sort out the problems with the new updates.
Maybe postponing the updates until the Stable update announcement forum topic reaches a state where a group of appointed people can give the signal that the update process got at a state where more stability is impossible with a rolling release distribution. We could make good use of these forum topics as they contain all the feedback from the community using the Stable channel.
I just realized that what I said above is exactly what happens with the update process:
Unstable -> Testing -> Stable
There is no need to create another level on this chain, maybe increasing the number of users on the unstable/testing channels would help.