Can upgrade from 15.1 to .2 be done by editing repos and running zypper in tty?

English Install/Boot/Login
1

Folks:

Haven’t seen any problems showing up on the forum for 15.2 so far . . . I have 15.1 installed on a 09 MBPro and was using MATE until the recent problems . . . and then even after the fix MATE isn’t back to what it was . . . I’m thinking about just jumping up to 15.2 . . . .

Is it possible to edit the repos where it says “15.1” to say “15.2” and then running zypper to upgrade the system . . . or, nope that isn’t “released” yet so the only way to get there is by booting the install as disk or USB drive and either running a fresh install or would that offer “upgrade” and most of the present install would be generally preserved as far as grub and so forth goes??

2

Leap 15.2 is still in alpha testing. I do not recommend upgrading to that.

I am testing 15.2 in a virtual machine. But I won’t upgrade anything or install on a real machine until it is closer to being ready.

3

@nrickert:

Thanks for the reply, I do remember that was your previous advice . . . is it something that could or would do sabotage to the host machine, or it’s just kinda spotty in what works or doesn’t work?

I wanted “stability” or less maintenance than TW for my laptop, that’s why I went with 15.1 in it, but then MATE got all busted up and it still is “wonky” as far as toolbars go . . . browser windows disappear to . . . somewhere . . . had to log out of MATE and go to Gnome Classic to find all of the opened windows, etc.

So, would 15.2 be “worse” than breaking a pattern functionality?? I’m not entirely in a raging rush, but I did find the current build of 15.2, but it would take some fiddling to get it ready to install as right now the Nvidia drivers aren’t providing access to a working TTY . . . it isn’t exactly full functioning in what I have now . . . so I was just going to jump up to “something completely different”???

I might have a version of OpenBox set up somewhere in one of my installs . . . but I usually do “bare metal” and if they make a mess I wipe, etc . . .

4

If you’re familiar with virtualization like Virtualbox,
Try installing MAT as your original DE… Without any other DE on the system (I don’t have to repeat how to do that, right?)
If it looks right in a VM, it’s highly likely that means that it should look that way on a real, physical install as well.
Then, you can decide… If your machine is very new with relatively little installed and configured then it might make sense to just re-install.
At least, that might be a more straight-forward approach without headaches instead of trying to track down exactly what might be causing your MATE problems.

TSU

5

@tsu2:

Thanks for the suggestion . . . since there isn’t a “live” version, the idea of installing into a VM is a good idea . . . . And, nope, you don’t have to repeat the “bare bones install” instructions . . . I think I have that one under control . . . . : - )

6

Sure. It differs from what TW users routinely do only in that editing first isn’t needed. I’ve done at least three 15.1 to 15.2s that way so far, but they are all on test PCs. As the others have written, it’s not a good plan for an only installation that needs to be reliable daily.

7

Have you tried creating a new user account, and running MATE on that? This tests whether your problems are due to messed up user settings.

8

I use MATE as well and this might help.

I found that there is something that changes the access to the user’s dconf/user file to root:root and that cause mate-panel to die.

I caught it by having the journalctl log running in a terminal and saw that mate-panel could not access it - after about a minute mate-panel and the interface went south.

my temporary fix is a root cronjob once a minute to check the file in the box - here the user is user (if user is 1000) this will keep mate-panel from going south. No idea what changes it to root:root.

 /usr/bin/chown 1000:100 /run/user/1000/dconf/user
9

@mrmazda: Alright!! Thanks for that reply . . . that might be the “easy” way to upgrade . . . most of my linux installs are “test” “mules” . . . only problem right now is no access to the TTY, have to change out of nvidia drivers to get back to that.

@nrickert: Haven’t done that, in that we know that some of the mate packages got broken . . . then were apparently fixed?? So, you are suggesting that something might have affected the original user account in the breakage/repair so the only way out of that is by making a new account? Easy enough to check . . . .

@nrickerts

@larryr:

OK, well similar answer . . . the mate panel and a couple packages relating to it were done broke . . . but in my case the repairs didn’t seem to do a full repair . . . so possibly there are other damages . . . similar to what nrickert is referring to via the user account . . . that you found could be compensated for by your cronjob . . . ??

@et al: Thanks for the responses, it’ll take some time to check out these ideas . . . .

10

@nrickert:

So I did manage to create a new user account and logged into MATE session and it appears to be working as it should, at least the original MATE drop down menu from the panel is working as it did before the recent breakages . . . . So how would I try to fix the original user account so that MATE would be able to recover itself?? Possible? Or if I want MATE I have to use the new account??

n_s

11

Try emptying cache. Log out of your normal user account and delete all content from ~/.cache/. It sometimes fixes Plasma, so might work for other DEs.

12

@mrmazda:

OK, thanks for the reply . . . not exactly clear how to do what you are suggesting . . . so this would be in a TTY? Or, I would be in the new user account and I would run some command in the terminal? Or, new user account but i would navigate to the directory of the problem account, open the “cache” and then drag stuff to the trash???

I have MATE again, but still having problems with getting into TTY in many of my Nvidia driver systems . . . just jumping from one problem to the next, needed Nvidia to get “suspend” function, etc . . . . So if there is some command to “delete content” . . . or is this an “sudo zypper rm ~/.cache” command that wipes the directory and everything in it and then on reboot a new ~/.cache is created???

13

Ctrl-Alt-F3 from the login greeter. Login as the problem user.

rm -R ~/.cache/*

Log out. Alt-F7. Login normally.

14

@mrmazda:

Thanks very kindly for that, I’ll give it a try in a bit . . . have to jump out of the computerland for awhile.

15

@mrmazda:

Arighty, did try that, but the nvidia drivers for this '09 computer are not letting me get to a TTY . . . I guess I could use the GUI terminal from the working MATE session to somehow point to the cache of the not working as well user account cache???

16

If you did that there should be open files there, so the result wouldn’t be empty.

Boot something else, mount the /home filesystem, delete the cache, boot normally.

17

@mrmazda:

Well, in this computer the Leap system is the only linux install . . . I have a persistent live usb drive with Lu 19.10 loaded . . . still somewhat beyond my working knowledge on “mounting” stuff . . . and deleting stuff in that mounted filesystem . . . . It’ll again be some time before I can get the time to play with that one.

18

So continuing along this thread, since the original user’s MATE DE remained damaged I went ahead and ran an “upgrade” from usb drive to bring the system up to Leap 15.2 . . . did that about a week ago. On reboot the newest kernel was 5.3.0 xxx . . . and the system shows an “error” face saying there “is a problem” and to “log out” . . . . Logging out by hitting the return key, as no mouse cursor is working, brings the error window again.

Shutting down and rebooting into recovery does not work, but finally dropping back to what was likely the Leap 15.1 kernel 4.12xxx . . . and the system boots up fine, have to log into the new user account to get MATE working . . . but it runs fine with 4.12.

This morning I booted up a persistent live usb drive with Lubuntu & 5.2 kernel and the system boots and runs . . . I rebooted into Leap 15.2 w/ 4.12 kernel and ran zypper upgrade and it showed 1776 packages to upgrade, including a new 5.3.1xx kernel for a total of 1GB to download . . . was on a slow internet connection, so an hour later I was rebooting and, same deal, the two 5.3 kernels bring error, “log out” window . . . but dropping back to the next choice of 4.12 and I’m back into a GUI and typing this post.

Not sure if this '09 MBPro is now “too old” for 5.3 kernel, or whether there is some “dependency” issue(s) with 15.2 that are causing a problem. After I did the initial install and had the problems I checked YAst for “dependency issues” and it showed “clean” . . . in ubuntu we can run the “sudo apt -f install” command to try to repair something and often that actually works . . . here trying to boot “recovery” doesn’t seem to get past the “error-please log out” window.

I realize @nrickert said it was “too early to do a 15.2 install” . . . but for the experiment I did it . . . . So far it is only running on the 4.12 kernel, which I guess I would need to ask Yast to keep it install before running another kernel upgrade?? as likely the newer kernels are “not recognized”???

19

Edit “/etc/zypp/zypp.conf”. Look for the line:


multiversion.kernels = latest,latest-1,running

and change that to read


multiversion.kernels = oldest,latest,latest-1,running

The oldest kernel will then be kept. So that should keep your 4.12 kernel from going away.

I don’t know why your 5.3 kernel is causing problems. It is working fine here on my test system (in a virtual machine). But perhaps it is a graphics card issue.

20

@nrickert:

Thanks for the reply, appreciate that information . . . but, yep, somewhat “odd” . . . as I do see the “nvidia” graphics splash page go by, at least in the 4.12 kernel boot, if not also in the 5.3 kernel selection . . . . But also in today’s downloads I did see “nouveau” mentioned as well as “mesa-dri” . . . I’m pretty sure I had “nvidia” repos added in Leap 15.1 so would that “remain” as part of the “upgrade”???

I recall seeing when the upgrade began they removed all of the repos from 15.1 . . . but then I don’t recall if I saw them refill all of the repos . . . ??? Well, I’ll try to hold onto this 4.12 kernel and then run a few more zypper upgrades and see how that transpires . . . in the nearing future.