X is Borked - Upgrade?
As it says, my X setup seems nicely borked. I had executed some scripts in /etc/X11/xinit. I was looking at installing Mate instead of XFCE. That screwed things.
Now I run 'startx' from runlevel 3 and I get a blue screen with one grey icop in top left & top right corners and this error Quote:
I deserve this because I had been holding on to stuff in /etc/X11 for years simply to keep the look & feel of XFCE that I was used to, so I know better than to try and fix this, so a clean install of current is coming, I want Mate, DEFINITELY not KDE, and not XFCE. What's my kosher way of doing that? Do I vaguely remember some script? Has anyone tried it? I can afford to, because I have a working backup to get me back to where I was recently. Otherwise, with the standard iso, if I leave out kde & XFCE, and put Mate there instead, does it work? |
kosher-schmosher... my favourite install method is to grab Eric's liveslak ( in this case the mate variety) and then run setup2hd
From Eric's wiki: Quote:
That's how I've done it in the past, and I consider it failsafe. |
Brilliant. Thanks for the reply.
Updating my homedir backup at the moment. I'm impressed with these LiveSlak USBs. I was on the Cinnamon one this morning and even got Libreoffice printing in short order. I'll choose carefully what to restore. I see my homedir has sprouted an fvwm directory, for instance. No doubt I'll manage to make an adventure of it but I have the backup. |
When I installed current on this box, I used Eric's liveslak, and it was completely no-drama. :) (If another testimonial is needed.) he's got it set up to use ventoy now, too.
|
I wouldn't go with the nuclear option just because (it seems) like XFCE isn't behaving. There's alot of fixes if you search for that error message, most have to do with starting under dbus-launch xfce4-session. Probably some start-up script is wrong.
|
I'm pretty sure nothing major is wrong, but finding what variables are jammy side down promises to be a nightmare, given that I've been copying over/etc/X11/* for some years into each new install. That will include /etc/X11/xinit/. I have a complete backup, and I'm going to be very fussy that only needed data goes back into the homedir. As it was, I couldn't set runlevel 4, but just ran startx from runlevel 3.
If I don't like it, I can just go back to earlier days. I even have a more historical backup. |
Well, I ran the setup2hd script, copied over the packages very nicely, and left me in the lurch in the final setup. Probably my mistake, but I know how to do that stuff anyhow and I did it. So I'm on a new 64bit ext4 filoesystem for /, the 6.6.10-huge kernel, no initrd, and runlevel 3.
It boots fine, but can't mount / because of errors. It asks me to enter the root passwd. But without / mounted, it can't check a passwd. I did try 'root' and no passwd. root was rejected, and no passwd was a reboot command. So I booted on the live usb to fix it. There was no errors! I checked all partitions, and only /home had minor errors. Is there a root passwd hardcoded into busybox in the ramdisk? What is it? I'll go back and try making a ramdisk and using the generic kernel. But I shouldn't have to - that's the whole point of the huge kernel, isn't it? EDIT: Made the ramdisk and booted it. It still gives me this root disk error on a new / filesystem. I even booted debian and tried the e2fsck there. No joy. |
Can you finally please please please be actually specific when you state things like "left me in the lurch in the final setup", "I know how to do that stuff anyhow and I did it" "because of errors" and all the other similarly vague remarks you made over time.
There is nothing in your posts that would actually entice me to hand you something relevant. I would love to help, but not like this. |
Quote:
Code:
*********************************************************** Quote:
Quote:
Quote:
|
Yes, what you showed was the error message I got. Let's be clear, the / partition is fine. I have checked everything, mounted all filesystems, checked all filesystems, and done my best to trap all errors going through. That involved half a dozen reboots. Here's what happens.
I didn't get to the end of setup2hd, but was dumped out. Using the live usb, I mounted the future directory tree and chrooted into it, I set up the root passwd before I posted, so obviously I had exited setup2hd before that point. Now I got in, and could progress. I ran 'init 3' and it finished and brought me to a login. Because I remounted / rw and ran 'init 3' it ran rc.M and I got some logs. dmesg was useless. Here's /var/log/syslog in it's entirety: Code:
Mar 22 17:41:10 darkstar kernel: Speculative Return Stack Overflow: IBPB-extending microcode not applied! So now I'll download the latest live iso and run some selective updates, particularly kernel and modules. |
Quote:
|
It's not the disk, to my mind. This stuff is less than 2 years old and not stressed. I checked the disk as a matter of form. What does happen is that the usb disk fills it's read or write cache and never thinks to stop and sync itself. At that point, programs are capable of doing funny things. That's my guess. I've been sitting with a terminal open running sync during setups. But I didn't get a chance to interrupt the sxz archives to nvme transfer.
Anyhow, current has moved on. The upgrade will be Sata --> nvme, which I will do in sections and I reckon that will work. It will also give me a different kernel |
I'm marking this solved :D.
I'm online, & three quarters sorted. The "/sbin/e2fsck is a directory" error proved to be missing files in /etc. Perfectly obvious once you think about it - NOT! It's madness, actually. You know, after an install, there's a trialsome routine of setting up basic config files. Well, I had wiped / clean, and setup2hd had turfed me out with skid marks on my backside, so those basic files in /etc/ were never set up. So I got back in, ran init 3, and selectively copied over one file at a time - /etc/fstab, /etc/hosts, and other configs I has set up to my liking, and knew to be unpolluted by earlier versions. Now I'm starting in runlevel 4. It starts the last window manager I had, and I can select Mate, XFCE, or some half-assed spartan thing. They coloured the background, threw in a disgusting terminal, and called it a window manager. It has the hallmarks of an alcoholic CS student blown out by his girl friend. I hadn't seen that option of choosing desktops before. I could have been locking myself out of it by copying over /etc/X11. Mate is set up to look much the way I had XFCE set up previously. The only thing I lack in Mate is a keyboard shortcut to switch between the 4 screens or windows. XFCE used to do it with Ctrl_F1-F4. What does Mate use? Current is from this week, Mate is from last month. To do is: installing from Alien Bob's repo; the task of setting up look 'n' feel stuff in Mate's Labyrinthine menus; Building or installing Slackbuilds. |
OK, so the sddm display manager (the login screen) is driving you nuts. You can get rid of The virtual keyboard:
edit line 24 in /etc/sddm.conf so that it reads InputMethod= and nothing else. If you prefer the old xdm login screen, comment out the sddm lines (lines 43-44) in /etc/rc.d/rc.4 For window switching, look for the menu item Keyboard Shortcuts or run mate-keybinding-properties from terminal. |
Sorry I was unclear.
Window maker is the half-assed window manager I was bellyaching about. I'll tolerate the login screen for the moment, but mark that spot you mentioned with a comment. I think I know how to get rid of Window maker. Experience teaches me to stop fixing something at the earliest opportunity, and keep your work reversible where possible. SDDM, is it? I hadn't run that before. Thanks for the mate tuition. Those keybinding properties are definitely future work. I'd rather leave them as they are and learn them than have to fart about every time I upgrade. I'll configure Mate at my leisure. |
All times are GMT -5. The time now is 06:39 PM. |