pc-kernel umount failure

[Sergio Costas]

Doing some install tests with the Oracular October 8 ISO, I had an installer crash.

I booted with a live system and proceed to wipe the internal SSD hard disk using `blkdiscard -f /dev/sda` (this disk is quite old, so it's recognized as a SATA one, not as a NVME). After that, I completely shut down the computer, and booted the Oracular Desktop LiveCD using an USB stick created with the October, 8 ISO. I chose "Spanish" and "Spanish keyboard", no `accessibility` options, connected to my WiFi (no Ethernet available in the laptop), clicked next, next, chose `extended selection` to install "all the stuff", clicked next, marked both `Install third party software for graphics and WiFi hardware` and `Download and install support for additional media formats`, and when I press the `Next` button, the installer crashed.

I attach the contents of the `/var/log` folder.

I also checked the first 400KBytes of the hard disk, and it's all zeros. No data there, as expected after a `blkdiscard`.

The laptop is an Asus Zenbook UX21E from 2012, with 4GBytes of RAM, with an Intel Core i5-2467M CPU.

The bug can be consistently reproduced after using `blkdiscard`. But if the disk has partitions, it doesn't seem to fail.

I didn't test with other languages. Also, didn't test using `normal selection` or not marking `third party drivers` or `additional media format`. Only tested that specific combination.

No FDE/TPM install is being done (the laptop is too old for that).

[Sergio Costas (rastersoft-gmail)]

I'll keep the crashed computer on for today, in case you need more files to diagnostic this.

[Olivier Gayot (ogayot)]

FWIW bug 2083952 and bug 2075994 have the same symptoms

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Olivier Gayot (ogayot)]

In all the reports I've seen so far, there's a cancellation during `_get_systems` when we're in the middle of requesting info from snapd though its API:

```python
return await self.app.snapdapi.v2.systems[label].GET()
```

However, cancelling on the client (i.e., subiquity) side does not necessarily means the server (i.e., snapd will stop executing its query handler). So we can't immediately expect snapd not to be using the systems directory anymore. We'd need to wait until the request has been responded.

```python
        self._examine_systems_task = SingleInstanceTask(self._examine_systems)
```
^ this is created with `cancel_restart=True` (the default) and is probably the reason for cancellation?

```python
        self.app.hub.subscribe(
            (InstallerChannels.CONFIGURED, "source"),
            self._examine_systems_task.start_sync,
        )
```

[corrado venturini (corradoventu)]

Same problem installing from Ubuntu 25.04 "Plucky Puffin" - Daily amd64 (20241201)
I had the problem twice with the same usb pen, the third time, waiting for the activity on the disk to finish, the installation completed well.

[Sebastien Bacher (seb128)]

@corrado, is that specific to some options? or do you just pick the default for everything?

[corrado venturini (corradoventu)]

Install in English with Italian keyboard, interactive install, extended selection, 'Install third party software...` and `Download and install support...`, press the `Next` ... the installer crashed.
I may ?easily? reproduce the problem because it happens very often, so tell me i may produce some more useful info.

[corrado venturini (corradoventu)]

Here the ubuntu report from the successful install:
{
  "Version": "25.04",
  "OEM": {
    "Vendor": "Gigabyte Technology Co., Ltd.",
    "Product": "H510M H",
    "Family": "H510 MB"
  },
  "BIOS": {
    "Vendor": "American Megatrends International, LLC.",
    "Version": "F14"
  },
  "Arch": "amd64",
  "HwCap": "x86-64-v4",
  "GPU": [
    {
      "Vendor": "8086",
      "Model": "4c8b"
    }
  ],
  "RAM": 15.4,
  "Disks": [
    500.1,
    500.1
  ],
  "Partitions": [
    32.7,
    0.5
  ],
  "Screens": [
    {
      "Size": "600mmx330mm",
      "Resolution": "1920x1080",
      "Frequency": "59.96"
    }
  ],

[Ubuntu QA Website (ubuntuqa)]

This bug has been reported on the Ubuntu ISO testing tracker.

A list of all reports related to this bug can be found here:
http://iso.qa.ubuntu.com/qatracker/reports/bugs/2084032

[corrado venturini (corradoventu)]

Today i had again the same crash but may be I was able to get documentation of the problem.
Attaching log and crash

[corrado venturini (corradoventu)]

Attaching also some screenshots

[corrado venturini (corradoventu)]

Note: same USB key, same PC, same partition successful install with the same options just wasting time during installation to take screenshots of all pages.
I had the same experience with preceding ISOs