Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
CaptainSarcastic
Jul 6, 2013



wormil posted:

Linux baby here...

I have an old laptop and have been installing various distros for fun and all was well until I installed the 3rd. I made 3 partitions and installed Mint on the first, Peach on the second, (all was good) and then Pop! on the third and now I can only boot Pop!. My guess is that I unintentionally changed the boot drive? There is a 4th partition with the original grub file and it seems to be intact but isn't being read at boot. I was tempted to delete the Pop partition but there is probably an easier fix. I have a follow up question but will save it until later.

How is the disk partitioned? My guess would be the first two installed GRUB to EFI and then Pop installed an MBR bootloader, or vice versa. Does Pop have a bootloader that displays before booting the OS? It might be possible to edit it to check for other OS installs and boot from it.

Adbot
ADBOT LOVES YOU

BlankSystemDaemon
Mar 13, 2009



Storm One posted:

gksudo is not part of Debian anymore, so no.

EDIT: fixed by rebooting :doh: apparently
I don't understand how Windows users' favorite method of masking problems becoming completely commonplace on Linux doesn't terrify people.

RFC2324
Jun 7, 2012

http 418

D. Ebdrup posted:

I don't understand how Windows users' favorite method of masking problems becoming completely commonplace on Linux doesn't terrify people.

Its a desktop. Frequently, a problem is transient, and a reboot, which should restore things to the sane defined configuration, fixes them. And again, its a desktop, who cares if it stops booting, the data should be easily recoverable

It it happens again, dig deeper. If its a server, figure out what is happening before you do anything, and if you know what's happening you can fix it without a reboot.

xtal
Jan 9, 2011

by Fluffdaddy
Probably had to restart polkit? It's the daemon pkexec uses right? Clearly I have not caught up since gksudo though.

wormil
Sep 12, 2002

Hulk will smoke you!

CaptainSarcastic posted:

How is the disk partitioned? My guess would be the first two installed GRUB to EFI and then Pop installed an MBR bootloader, or vice versa. Does Pop have a bootloader that displays before booting the OS? It might be possible to edit it to check for other OS installs and boot from it.

[SOLVED]
4 partitions: 1 MBR (leftover from the original Windows installation), and 3 Linux filesystems
Apparently this is a thing when dual booting Ubuntu and Pop as they use different bootloaders. The solution took more time to find than execute, run os-prober and update-grub.

Also, as an aside if anyone is interested. I've tested 7 distros on this old laptop and Peach/Patriot is the snappiest, not measured, just based on use. Peppermint and Mint are very close seconds, Pop! is a very close third. Any of those four are fast enough for basic stuff. Ubuntu and Minjaro were faster than Windows but sluggish and would be mildly annoying in use unless you've never used anything better. Also tested Chromium but that was before I fixed the laptop overheating so it's not fair to comment but I believe it would be fine if you want a google ecosystem.

Storm One
Jan 12, 2011

xtal posted:

Probably had to restart polkit? It's the daemon pkexec uses right? Clearly I have not caught up since gksudo though.

Maybe? Earlier in the day, I did have to puleaudio --start because I had no sound, so that's probably it.

In hindsight, I should have started by reporting that this happened after moving the root drive from one computer to another and expecting it to Just Work (which tbf it mostly did) and having to install AMD firmware from nonfree because of the GPU in the new (actually older) motherboard. All of which also happended to coincide with the day Debian updated from 10.5 to 10.6.

Mr Shiny Pants
Nov 12, 2012
I did something dumb, I did an apt uprade while running my VM and the system died while upgrading.

Now Grub does not know what to start, I just get a shell.
What is the most foolproof way of fixing this?

NihilCredo
Jun 6, 2011

iram omni possibili modo preme:
plus una illa te diffamabit, quam multæ virtutes commendabunt

wormil posted:

Also, as an aside if anyone is interested. I've tested 7 distros on this old laptop and Peach/Patriot is the snappiest, not measured, just based on use. Peppermint and Mint are very close seconds, Pop! is a very close third. Any of those four are fast enough for basic stuff. Ubuntu and Minjaro were faster than Windows but sluggish and would be mildly annoying in use unless you've never used anything better. Also tested Chromium but that was before I fixed the laptop overheating so it's not fair to comment but I believe it would be fine if you want a google ecosystem.

Did you choose XFCE for all of those distros? Otherwise it's not surprising that an XFCE distro (Peach / Patriot) feels faster than those that default to GNOME.

RFC2324
Jun 7, 2012

http 418

Mr Shiny Pants posted:

I did something dumb, I did an apt uprade while running my VM and the system died while upgrading.

Now Grub does not know what to start, I just get a shell.
What is the most foolproof way of fixing this?

Do you still have install media?

Because reinstalling is going to be the preferred way to fix it, and plan B still involves boot media

xtal
Jan 9, 2011

by Fluffdaddy
Wrote up a whole wrong post here because I made too many assumptions about the problem.

Do you see an error message? Is it a shell in grub or in Linux?

xtal fucked around with this message at 16:26 on Sep 28, 2020

RFC2324
Jun 7, 2012

http 418

xtal posted:

Wrote up a whole wrong post here and edited to this.

Do you see an error message? I initially thought you were describing a problem with the bootloader, but it sounds like the bootloader is working if it can drop you to shell. The problem is in the startup process or initramfs.

Modern bootloaders have a built in shell, I assumed that's what he meant.

Bad assumption on my part

xtal
Jan 9, 2011

by Fluffdaddy

RFC2324 posted:

Modern bootloaders have a built in shell, I assumed that's what he meant.

Bad assumption on my part

Yes, I edited my post further once I realized this. I'm gonna shut up for a minute.

RFC2324
Jun 7, 2012

http 418

xtal posted:

Yes, I edited my post further once I realized this. I'm gonna shut up for a minute.

Bootloaders are a mess, and we are just guessing based in a crumb of info off the loaf of the problem.

I'm not going to stop because it's fun, and blind troubleshooting like this is a useful skill when you get dropped into an outage where restoring service asap is the most important thing

Mr Shiny Pants
Nov 12, 2012
Sorry for not giving more info, I just get dumped to the GRUB shell.

From there I can do an LS and so forth to get the installed disks and what have you. The problem is that I have quite a lot of disks in the system making it hard to guess which one I need.
I'll try and boot a Live CD and get the correct disks first.

When I try to boot vmlinuz it says I need to boot a kernel first. This is not really my forte and I just want it to work again. :)
I'll try the Boot Repair Disk and see what that does.

wormil
Sep 12, 2002

Hulk will smoke you!

NihilCredo posted:

Did you choose XFCE for all of those distros? Otherwise it's not surprising that an XFCE distro (Peach / Patriot) feels faster than those that default to GNOME.

XFCE Minjaro was the most sluggish, very slightly worse than Ubuntu, which surprised me since people brag up Minjaro as fast on old systems. Mint Cinnamon is Gnome based and extremely similar to Peach in performance. I don't believe that the desktop is the only difference in speed, not that there is anything wrong with that, on a modern system they would be indistinguishable in performance. Minjaro did seem to boot faster, and Mint seems to boot slowest but again I didn't measure anything and it's a completely subjective opinion.

Pop! IMO was by far the most attractive interface but it's a bit bossy and the clean layout takes a smidge longer to navigate. Peach is the most cluttered desktop by default and has the ugliest wallpaper but it can be customized. Mint is a bit utilitarian but intuitive and easy to use. Peppermint was similar to Mint, might have been faster, less attractive, it didn't click with me and there was some driver issue (no doubt easily solved), so I spent very little time with it. In the end, Mint and Peach are tied with me. When using Peach I barely notice the UI but Mint is a tad more familiar and intuitive to a Windows user.
(Note, I have never used MacOS X and can't say how well Peach simulates the experience.)

wormil
Sep 12, 2002

Hulk will smoke you!

Mr Shiny Pants posted:

Now Grub does not know what to start, I just get a shell.
What is the most foolproof way of fixing this?

Is it possible the solution to my problem would work for you? 2 posts above yours.

Os-prober run from boot disk.

Mr Shiny Pants
Nov 12, 2012

wormil posted:

Is it possible the solution to my problem would work for you? 2 posts above yours.

Os-prober run from boot disk.

Thanks, I'll give it a shot.
Edit: Just downloaded the KUbuntu 20.04 installer, Rufus is writing it to USB stick and we'll see from there. :)

Edit2: os-prober found my two OS's on the NVME drives but I couldn't get it to upgrade GRUB via the Live CD, which seems pretty normal seeing that it booted from USB. Missing the required knowledge I just added the Boot-Repair PPA and had it do the "quick repair" thing and lo and behold my system boots again.

Still something in journalctl about it not starting libvirtd or something. That is probably also the reason the system crashed in the first place. Something seems to have changed.....

Edit3: running sudo dpkg --configure -a and sudo apt upgrade and rebooting seems to have cleared things up. Libvirtd starts up and everything looks normal.

Bit weird.

Mr Shiny Pants fucked around with this message at 21:10 on Sep 28, 2020

RFC2324
Jun 7, 2012

http 418

Mr Shiny Pants posted:

Thanks, I'll give it a shot.
Edit: Just downloaded the KUbuntu 20.04 installer, Rufus is writing it to USB stick and we'll see from there. :)

Edit2: os-prober found my two OS's on the NVME drives but I couldn't get it to upgrade GRUB via the Live CD, which seems pretty normal seeing that it booted from USB. Missing the required knowledge I just added the Boot-Repair PPA and had it do the "quick repair" thing and lo and behold my system boots again.

Still something in journalctl about it not starting libvirtd or something. That is probably also the reason the system crashed in the first place. Something seems to have changed.....

Edit3: running sudo dpkg --configure -a and sudo apt upgrade and rebooting seems to have cleared things up. Libvirtd starts up and everything looks normal.

Bit weird.

can't tell you the cause of the initial crash, but the rest happened because you had an interrupted update process.

Mr Shiny Pants
Nov 12, 2012

RFC2324 posted:

can't tell you the cause of the initial crash, but the rest happened because you had an interrupted update process.

I figured, the problem was that Libvirtd crashed after updating something. Don't know why. It did it a second time after coming up clean though. I don't have a weird configuration.

xzzy
Mar 5, 2009

linux world: persistent device names are super cool and good, use, them.

linux: enp73s0f3u1u2c2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

:hmmyes:

RFC2324
Jun 7, 2012

http 418

xzzy posted:

linux world: persistent device names are super cool and good, use, them.

linux: enp73s0f3u1u2c2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

:hmmyes:

Have you rebooted to see if its a persistent name? Just because its a dumb name doesn't mean it doesn't stick

xzzy
Mar 5, 2009

It's 100% persistent, the code is doing exactly what it's supposed to do. That line noise of an interface name is just hilarious to me.

RFC2324
Jun 7, 2012

http 418

xzzy posted:

It's 100% persistent, the code is doing exactly what it's supposed to do. That line noise of an interface name is just hilarious to me.

:agreed:

The dumb rng names make modern linux better because reasons, maybe one day I will understand.

It was hilarious when it was a different rng name every time an interface was initialized tho

hifi
Jul 25, 2012

just use 'ip link altname'

BlankSystemDaemon
Mar 13, 2009



On FreeBSD, persistent device names are handled via device wiring through hints.

other people
Jun 27, 2004
Associate Christ

xzzy posted:

linux world: persistent device names are super cool and good, use, them.

linux: enp73s0f3u1u2c2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

:hmmyes:

Oh come on. That's a usb nic. It is supposed to suck.

CaptainSarcastic
Jul 6, 2013



I gave up a long time ago trying to keep track of the changing naming conventions. My wireless connection is wlp60fbatun3c now? Fine, whatever. I can just barely manage to keep track of sda, sdb, and so on. The rest of it I ignore unless for some reason I absolutely have to mess with it.

other people
Jun 27, 2004
Associate Christ
It is all pretty well documented:

https://systemd.io/PREDICTABLE_INTERFACE_NAMES/

https://www.freedesktop.org/software/systemd/man/systemd.net-naming-scheme.html



The names are certainly nicer if your system provides slot/port details via smbios or acpi or whatever. This is pretty common on enterprise class stuff but not so much on home hardware.

If your system only has one NIC then by all means boot with net.ifnames=0 and you'll be back to your single interface always being eth0. But of course if you ever add a second NIC you'll surely want to have predictable names back on.


USB is an unfortunate case as there is just no "neat" piece of info for the udev netid helper to use to consistently generate the same name for a given USB device. So you get a goofy name like enp73s0f3u1u2c2 which describes that nic port's location on the usb bus.

RFC2324
Jun 7, 2012

http 418

other people posted:

If your system only has one NIC then by all means boot with net.ifnames=0 and you'll be back to your single interface always being eth0. But of course if you ever add a second NIC you'll surely want to have predictable names back on.

I didn't realize you could go back, thats neat!

Too bad I haven't seen a linux box that didn't have some psuedo interfaces as well as the physical ifs in years

other people
Jun 27, 2004
Associate Christ

RFC2324 posted:

I didn't realize you could go back, thats neat!

Too bad I haven't seen a linux box that didn't have some psuedo interfaces as well as the physical ifs in years

Physical Ethernet interfaces are named ethX by the kernel. Then the new device announcement hits userspace via udev and some rule there renames them. Setting net.ifnames=0 turns off the systemd-udev built-in renaming on the userspace side. Custom udev renaming rules would not be affected by it.

Virtual interfaces such as vlans and bonds and tuntap and whatever have different naming behavior and standard systemd udev rules do not touch them.

So if your box has only one physical network port (or VM only has one virtual "physical" port) then you can boot with net.ifnames=0 and it doesnt matter how many taps or bridges or vlans or teams or whatever you have.

RFC2324
Jun 7, 2012

http 418

Ok, that becomes useful. Thanks

BlankSystemDaemon
Mar 13, 2009



Is it documented in a manual page? If not, then it's not documented.

other people
Jun 27, 2004
Associate Christ

D. Ebdrup posted:

Is it documented in a manual page? If not, then it's not documented.

Is what documented? The second link I gave is an html copy of a man page.

Man I wish we had a bsd thread.

waffle iron
Jan 16, 2004
man 7 systemd.net-naming-scheme

That has been a man page for systemd since version 243 released in June 2019. I looked it up.

Mr. Crow
May 22, 2008

Snap City mayor for life
Are ACLs typically not read/implemented during IO in non std/gnu/POSIX applications? E.g. GNOME text editor or mousepad.

I am getting weird behavior with NFS4_ACLs despite the metadata looking right and in strace it doesnt appear to even query the ACL, though im not very familiar with the tool...

BlankSystemDaemon
Mar 13, 2009



I assume you're not using OpenZFS (ie. not Illumos ZFS or derivatives like FreeBSD ZFS or NetBSD ZFS), because NFSv4 ACLs are not implemented yet.

zerofunk
Apr 24, 2004
I am having an issue with Fedora booting new kernels and am slightly out of my depth. A month or so ago, I installed the amdgpu drivers. Since then there have been a couple new kernel updates automatically installed. These don't boot, but an older one that is still on the grub menu - I am assuming it was in use when I installed the amdgpu driver - still boots just fine. My assumption is that this is related... maybe some sort of kernel module thing? Beyond that I am really not sure how to troubleshoot it. And when I say it doesn't boot... that is maybe not technically accurate, but it just sits with a black screen and a flashing cursor in the upper left. No real indication of what it is doing. It will reboot after doing SysRq + B.

At this point my next step would be to just uninstall amdgpu and re-install it. Seems like there should be a better way though.

RFC2324
Jun 7, 2012

http 418

I'd be inclined to say its because you don't need to install amd drivers, as they are just built into the kernel now(not bullshit drivers like nvidia, but real deal provided by amd drivers)

Mr. Crow
May 22, 2008

Snap City mayor for life

D. Ebdrup posted:

I assume you're not using OpenZFS (ie. not Illumos ZFS or derivatives like FreeBSD ZFS or NetBSD ZFS), because NFSv4 ACLs are not implemented yet.

Correct. I think it might be a mergerfs issue but I'm not sure...? Details in the bug ticket here: https://github.com/trapexit/mergerfs/issues/836

Adbot
ADBOT LOVES YOU

zerofunk
Apr 24, 2004

RFC2324 posted:

I'd be inclined to say its because you don't need to install amd drivers, as they are just built into the kernel now(not bullshit drivers like nvidia, but real deal provided by amd drivers)

From what I can tell, you do still need to install them for OpenCL support. That was why I installed them. Otherwise I think you’re right and I was using everything just fine for a few months without them.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply