There are some weird things here you are describing.
First, I’d make sure the BIOS is absolutely up to date on the board.
Second, if you’re using an nvidia card, make sure you’re using the nomodeset option.
Here is one site that goes over that:
basically you just add to the line that says “quiet splash” and just replace it with “quiet splash nomodeset”
This should at least fix the purple screen thing and the loop problem.
As far as your drives that may be a motherboard thing which is why you might want to udpate the bios, if that’s not it, there’s probably a boot setting getting screwy.
I have indeed updated the bios to the latest settings, and I am using an nvidia card (2080). Looking at the linked post, how can I access & modify /etc/default/grub without being able to log in?
You can edit your grub lines at boot up.
there are a few different ways depending on the version you’re running, here is a good link that goes over some of the ways:
do an online search for editing grub at bootup using ubuntu to get details
okay the UUID error indicates the drives name is getting changed during install somehow. That is very odd. I have been able to work around this before by forcing the drive name, but I don’t want you getting that into the weeds (shouldn’t be necessary normally).
I’ve seen this sort of behavior on an old laptop that had a really old UEFI mode. I’d have to turn off UEFI to get it to install and it would break the UUID when it tried to boot back up. I eventually figured out to change the download media to nonUEFI and disable UEFI in the BIOS and then I was able to get this to work.
I’m wondering if perhaps your ubunut version has UEFI installer by default and your PC UEFI is crapping out. You could disable UEFI but not sure if that’s going to mess up your windows install.
It is probably best at this point to try a different drive as it’s possible to break your windows install as you mess with this more.
Should I be booting from the USB installer or the UEFI installer? I think I’ve tried both, but it would be good to know if there’s a more correct way.
On one of my many reboots and reinstall attempts I managed to get to the installer without any error messages or warnings. The NVMe drive was correctly detected but the installation process eventually stopped with the following messages https://imgur.com/dgDeQeb. Note that Win10 is on partition 1, nothing else appears with labels partition 2-4, partition 5 was the EFI and partition 6 was the target destination for the Ubuntu install. This partition structure remains even if I erase/format everything [except Win10], after partition 1 the next available is labelled as 5.
As an aside, if I can get Ubuntu to work I don’t mind so much doing a complete wipe of the Windows partition. In fact, it might be best if linux is the only OS that touches the metal with all other OS’s operating through VMs
EDIT: It looks like I’ve already messed with enough stuff to break my Win10 install. I’m not so attached to it since I only built the machine recently, but this might be a good opportunity.
If you’re going to wipe the drive, then just do that and delete all the partitions so ubuntu can set it up properly with the defaults. There shouldn’t be 6 different partitions so I think something is getting really whacky here.
If you still need data off the drive that can probably be recovered if you install on a different disk and then plug this one in once the install is completed. You can probably read the messed up partitions and possibly even rebuild the boot sector on that drive to get back into windows.
The thing I think that is screwing things up is having NTFS partitions but I can’t say for sure since I’m not there to inspect them all.
Installation was fine and the drive was detected. Reformatted the drive and merged all partitions into a single one. Upon reboot I was first stuck on a purple screen. Rebooting and setting nomodset I am now getting similar nvme IO_PAGE_FAULT errors as earlier https://imgur.com/0yY03Qt.
I got a 250gb Samsung 860 EVO. I unplugged/disconnected every other drive (NVMe/HDD) and installation seems to work great. However, now at the log in screen (which I can consistently reach now, so that’s something!) I’m being immediately shot back to the log in screen after reaching the desktop for a few seconds.
It’s also worth noting that the log-in screen is extremely sluggish, so my guess would be for me to install nvidia drivers. That said, I’m not sure how to do so if I can’t reach the desktop.
Additionally, I go straight to the log-in screen without seeing a grub so I’m not sure where I could add the ‘nomodset’ parameter.
One thing to note is that during installation I was ‘spammed’ with what felt like hundreds of ‘PKCS#7 signature not signed with a trusted key’ messages at the end of the driver installation process. I forced a reboot, but everything seems to be working smoothly! (finally!)
Now to see if this can remain stable when I put back the nvme drive…
First boot: Stuck on purple screen for around 90s, then reach the log-in (everything working fine).
Second boot: Sent to grub (this hadn’t happened when the 860 evo was the only drive), then an nvme error message https://imgur.com/SGRfSnR then after about 90s sent to the log-in screen (everything works normally when logged in).
Third boot: See first boot.
Fourth boot: See second boot.
I assume the pattern continues in this way. I then formatted/reinstalled Windows on the NVMe just in case some old structure was getting in the way. After installing Windows successfully and rebooting into the Ubuntu drive (860 evo) I still see the above problem.
This is quite frustrating but at least its usable in a sense.
Nope scrape my comment you said everything works great in the previous post.
Is your NVME still with both Windows and Ubuntu ? maybe it boots in the old boot partition you made.
At boot up spam F8 ( usually this will open a menu to select a Storage device to boot from ) or Del or F2 ( the binds in most cases for your UEFI menu ) where you can choose to boot from the second ssd.
I have no problem switching between drives in the UEFI, since I’ve pretty much resigned myself to doing so with this system. However, it’s a little annoying to have to wait an additional 90 seconds when booting into the Ubuntu drive to reach the log-in.
It’s apparent at this point there is some issue with UEFI and NVME here.
As to how to fix it, that’s going to probably require more in depth knowledge than I can muster.
It certainly seems to me there is something off here about how the system is handling UEFI, but I’m not sure how the boot partition is getting created when using Ubuntu.
As you can see, EFI can be tricky depending on the platform. I have a feeling there is something here that can help, and if it were me, I’d install a very basic system using arch to figure out what was up (not recommending you do this as it’s a terrible rabbit hole).
I do think using EFI to switch is a bad option as it really degrades the experience, but hopefully you can use some google fu to find a workaround soon.
typical dual boot installation depends on grub or lilo boot loader being installed.
it one reason that linux distros recommend installing linux first (during partitioning set up an ntfs partition for windows) and install windows second
then install windows in that partition. grub or lilo will be edited to the os added to its files or you may have to manually edit it.
but i dont use dual boot so things might have changed since i went to linux.
But if you are installing on separate drives, during the boot process you can select the boot media by hitting the designated F-key and selecting the proper drive.
I think with everything that’s happened I’ll just keep the 860 EVO (maybe upgrade to 500GB), install Ubuntu on that, and just remove the NVMe drive altogether for this build.
The only purpose of Windows was to play games, but 1) I have a bad habit of distracting myself too much with games 2) I could learn more about setting up a pci passthrough using a Win10 vm.
Thanks again for your help! It’s really been comforting just knowing there was someone to talk to, even if it didn’t work out
Actually I’m using the exact same board, running into the exact same problem–Linux cannot detect my nvme drive whatsoever. So it seems to be a motherboard problem. There’s not much we can do now really, but when you upgrade your rig, be sure to change into a mb from a different OEM–that should solve the problem.