hard lock up

[jlittle]

Three days ago, my desktop started locking up, a complete freeze:

• no mouse movement
• ctrl-alt-f1 does nothing
• magic sysreqs do nothing
• "power" button does nothing
• reset (once I found it) or power off are the only options.

This is usually during the KDE login, or shortly after, but sometimes during the Linux boot (no quiet splash), and I've seen series of messages like "NMI watchdog: hard lock on CPU 4".

It's an Intel i7 6700, 16 GiB RAM, 250 GB Samsung 750 EVO SSD, on a Gigabyte H170-D3HP motherboard, UEFI boot. There's a WD 1 TB disk, and an old Seagate 500 GB disk. The SSD and the WD are formatted GPT, with most of both a btrfs partition. The old drive has the old MBR with several partitions from old installs. It's 2 years old.

I've tried all three of the kernels still on my system, from memory 4.13.0-37, 38, and 39. I've managed to run apt update and full-upgrade, and do a backup. Temperatures are reported as around 40 C.

I've run memtester and memtest86, that last for 13 hours, with no errors.

Is this possibly a borked Kubuntu problem? Or is definitely hardware? I'm not sure. It hasn't looked up in recovery mode, or booted from a systemrescuecd USB stick, but Kubuntu has run sometimes for as long as I've run those.

Any advice gratefully received.

[Snowhog]

Might be some helpful info at https://bugs.launchpad.net/ubuntu/+s...x/+bug/1596866

[GreyGeek]

Can you boot and run from a LiveUSB?

[jlittle]

Might be some helpful info at https://bugs.launchpad.net/ubuntu/+s...x/+bug/1596866

Thank you, snowhog, I had found that report, and read closely the posts, in 2016 and early 2017. Reading again, there's a new post dated 2018-04-04 (the page source, if I read it correctly, says it was 5 am UTC, which is yesterday for me) and mentioning a kernel I have. Promisingly, the report says with 4.15 the problem does not occur...

Can you boot and run from a LiveUSB?

I expect so, since my trusty systemrescuecd stick works. Downloading a Bionic daily now; I might be moving to 18.04 a little early.
I prefer release upgrades, but usually reinstall when an LTS comes round.

[GreyGeek]

Being able to run a LiveUSB is a good indication that the problem is software related and not hardware.

EDIT:
Correction! After I installed the x-plane demo I was given two options: exit or play the demo. I played the demo and it ran well. After 15-20 minutes I closed it. Later, I started x-plane from the executable. It hung while attempting to load scenery. I repeated it several times and got the same error. SO, it appears that 6GB of RAM is NOT enough to run X-plane, just like the requirements suggest.

[jlittle]

An update.

My 17.10 install stayed up from Thursday morning to Friday evening, so I sort of ignored the problem. But then, it started locking up during boot. So, I've installed bionic on the same btrfs volume, pretty much using oshunluvr's method, but with some foolishness on my part, leading to the grub rescue prompt.

I had some difficulty finding the daily 18.04 iso on a page with check sums. Most searches reached the released verions, and another a page with no sums. Eventually I got to http://cdimage.ubuntu.com/kubuntu/daily-live/current/. Just as well, because the check sums on first download didn't match. I repeated the download in exactly the same way, but the second was good. I used etcher.io to write the iso to a USB stick, as recommended by several here at KFN. Very simple.

I didn't want to overwrite the existing grub, so I installed with ubiquity -b.
In hindsight that was a mistake; the existing grub is on a dud install. It works fine, but updating it was problematic after adding a custom entry for bionic; but I could do that from recovery mode.

I like bionic beaver. plasmashell is crashing occasionally, but it restarts so fast there's little impact.

So now I'm working through my customization doc. I customize a lot, it's going to take me days.

I had strife with kwallet. I use kwallet as a password manager, and I have over a hundred passwords in it. To access a GPG encrypted wallet from a backup is a mission; you have to export and import the GPG keys first, and I failed to get the wallet into bionic by just copying files. Eventually I managed to run the 17.10 in a sort of safe mode, with VGA resolution, by booting in recovery mode, and fscking, and resuming the boot. Then I could export the wallet using kwallet. I needed to get at my firefox sync password.

After installing bionic and updating it, a snapshot was in order. I've been snapshotting my root volume with something like

sudo mkdir -p /mnt/snaps/root42
sudo btrfs subvolume snapshot / /mnt/snaps/root42

but gave an error saying '/' wasn't valid. I had to use

sudo mkdir /mnt/top
sudo mount /dev/sdc2 /mnt/top
sudo btrfs subvolume snapshot /mnt/top/@_bionic /mnt/snaps/root42

[GreyGeek]

You don’t snapshot @home_bionic

[jlittle]

You don’t snapshot @home_bionic

Yes, I did, something like

sudo btrfs subv snap /home /mnt/snaps/home

worked fine.

Regards, John Little