Bug 2458901 - An incomplete spanned btrfs makes anaconda not see the drive, and crash when rescanning
Summary: An incomplete spanned btrfs makes anaconda not see the drive, and crash when ...
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: python-blivet
Version: 44
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Vojtech Trefny
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard: AcceptedBlocker
Depends On:
Blocks: F44FinalBlocker
TreeView+ depends on / blocked
 
Reported: 2026-04-16 09:06 UTC by Kamil Páral
Modified: 2026-04-24 05:56 UTC (History)
12 users (show)

Fixed In Version: python-blivet-3.13.2-2.fc44
Clone Of:
Environment:
Last Closed: 2026-04-24 05:56:18 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Log file: /tmp/journal.log (693.93 KB, text/plain)
2026-04-16 09:06 UTC, Kamil Páral
no flags Details
Log file: /tmp/anaconda.log (11.62 KB, text/plain)
2026-04-16 09:06 UTC, Kamil Páral
no flags Details
Log file: /tmp/storage.log (104.85 KB, text/plain)
2026-04-16 09:06 UTC, Kamil Páral
no flags Details
Log file: /tmp/program.log (8.83 KB, text/plain)
2026-04-16 09:06 UTC, Kamil Páral
no flags Details
Log file: /tmp/packaging.log (10.70 KB, text/plain)
2026-04-16 09:06 UTC, Kamil Páral
no flags Details
Log file: /tmp/anaconda-webui.log (3.16 KB, text/plain)
2026-04-16 09:06 UTC, Kamil Páral
no flags Details
screenshot: no disks visible even if one is present (48.81 KB, image/png)
2026-04-16 09:15 UTC, Kamil Páral
no flags Details
screenshot: a crash after rescanning disks (41.71 KB, image/png)
2026-04-16 09:15 UTC, Kamil Páral
no flags Details
broken message during partition removal (84.67 KB, image/png)
2026-04-20 10:43 UTC, Kamil Páral
no flags Details

Description Kamil Páral 2026-04-16 09:06:19 UTC
Rescanning of the disks failed. org.fedoraproject.Anaconda.Error: failed to temporarily mount /dev/vda3 for btrfs operation

---[ System & Environment Information ]---
OS: Fedora Linux 44 (Workstation Edition)
Anaconda version: 44.30
Anaconda UI version: 68

Comment 1 Kamil Páral 2026-04-16 09:06:22 UTC
Created attachment 2137256 [details]
Log file: /tmp/journal.log

Comment 2 Kamil Páral 2026-04-16 09:06:25 UTC
Created attachment 2137257 [details]
Log file: /tmp/anaconda.log

Comment 3 Kamil Páral 2026-04-16 09:06:28 UTC
Created attachment 2137258 [details]
Log file: /tmp/storage.log

Comment 4 Kamil Páral 2026-04-16 09:06:30 UTC
Created attachment 2137259 [details]
Log file: /tmp/program.log

Comment 5 Kamil Páral 2026-04-16 09:06:33 UTC
Created attachment 2137260 [details]
Log file: /tmp/packaging.log

Comment 6 Kamil Páral 2026-04-16 09:06:36 UTC
Created attachment 2137261 [details]
Log file: /tmp/anaconda-webui.log

Comment 7 Kamil Páral 2026-04-16 09:14:37 UTC
Anaconda doesn't see a disk that contains an incomplete spanned btrfs filesystem (i.e. a btrfs filesystem that spans multiple disks, but one or more disks are missing). Such a disk doesn't show up in the Destination field, i.e. no disks are available for installation. If I use the Rescan Devices button, Anaconda crashes with the error from comment 0.

Reproducer:
1. Prepare a machine (UEFI VM in my case) with two disks
2. Perform a default F44 Workstation installation onto both disks. That means you select both disks in Destination, and select "Use entire disk". The resulting installation has btrfs spanned across both drives, which you can confirm by running `sudo btrfs filesystem show /`.
3. Remove the second drive from your system. The drive that remains contains ESP, /boot, and a / partition that contains the incompletely btrfs filesystem.
4. Boot F44 Workstation install media again, start Anaconda.
5. In Anaconda, the Destination shows "No disks available"
6. Hit Rescan Devices, Anaconda then crashes

Always reproducible.

Comment 8 Kamil Páral 2026-04-16 09:15:35 UTC
Created attachment 2137262 [details]
screenshot: no disks visible even if one is present

Comment 9 Kamil Páral 2026-04-16 09:15:59 UTC
Created attachment 2137263 [details]
screenshot: a crash after rescanning disks

Comment 10 Kamil Páral 2026-04-16 09:19:07 UTC
Proposing for a blocker discussion:
"The user must be able to select which of the disks connected to the system will be affected by the installation process. "
https://fedoraproject.org/wiki/Basic_Release_Criteria#Disk_selection
"The installer must be able to complete an installation to a single disk using automatic partitioning. "
https://fedoraproject.org/wiki/Basic_Release_Criteria#Disk_layouts
"When using the guided partitioning flow from the GTK-based installer or using the webui-based installer without entering the "Modify storage" page, the installer must be able to: Reject or disallow invalid disk and volume configurations without crashing." and "Remove existing storage volumes to free up space, at the user's direction."
https://fedoraproject.org/wiki/Fedora_44_Beta_Release_Criteria#Guided_partitioning
"The installer must be able to create and install to any workable partition layout using any file system and/or container format combination offered in a default installer configuration. "
https://fedoraproject.org/wiki/Fedora_44_Final_Release_Criteria#Disk_layouts

Comment 11 Adam Williamson 2026-04-16 15:13:58 UTC
Is the bug that installing to the incomplete set should *work*? Or that it should be cleanly rejected rather than crashing?

Comment 12 Kamil Páral 2026-04-16 15:46:41 UTC
(In reply to Adam Williamson from comment #11)
> Is the bug that installing to the incomplete set should *work*? Or that it
> should be cleanly rejected rather than crashing?

No, the bug is that I want to erase the whole drive with a fresh new default installation, and I can't. Because the drive is invisible to anaconda webui. 

I'm not saying that anaconda needs to be able to operate with a partial btrfs filesystem, that's unreasonable. I just want to be able to erase it.

Comment 13 Adam Williamson 2026-04-16 15:52:44 UTC
Oh, right, so like wanting to wipe a single disk from a RAID set. We've had proposed blockers like that before, I think, don't recall what we decided.

Comment 14 Vojtech Trefny 2026-04-16 17:06:35 UTC
(In reply to Kamil Páral from comment #12)
> (In reply to Adam Williamson from comment #11)
> > Is the bug that installing to the incomplete set should *work*? Or that it
> > should be cleanly rejected rather than crashing?
> 
> No, the bug is that I want to erase the whole drive with a fresh new default
> installation, and I can't. Because the drive is invisible to anaconda webui. 
> 
> I'm not saying that anaconda needs to be able to operate with a partial
> btrfs filesystem, that's unreasonable. I just want to be able to erase it.

Note, this is how this works with older versions of Fedora -- the disk is visible, it has a btrfs partition which can be removed, you just can't reuse the existing subvolumes (because you don't see these at all, because we can't get the information). I've found the blivet change that broke this and I'm working on a fix.

Comment 15 Adam Williamson 2026-04-16 20:34:18 UTC
Discussed at 2026-04-16 Go/No-Go meeting, acting as a blocker review meeting: https://meetbot-raw.fedoraproject.org/meeting_matrix_fedoraproject-org/2026-04-16/f44-final-go-no-go-meeting.2026-04-16-18.00.html . Accepted as a blocker as a violation of the criteria cited in comment #10 . Vojtech, if you could finish the fix and get a build done as soon as possible it'd be awesome, thanks.

Comment 16 Fedora Update System 2026-04-17 12:28:37 UTC
FEDORA-2026-985539cc51 (python-blivet-3.13.2-2.fc44) has been submitted as an update to Fedora 44.
https://bodhi.fedoraproject.org/updates/FEDORA-2026-985539cc51

Comment 17 Geraldo Simião 2026-04-17 13:38:19 UTC
I tested the new python-blivet-3.13.2-2.fc44 build on a VM used before to confirm the bug, and now the anaconda webui identified the device right at the first try. I tryed to rescan devices too to see if the crash happens but it found the device correctly. so, as long as I can tell, this build fix the issue.

Comment 18 Fedora Update System 2026-04-18 01:32:08 UTC
FEDORA-2026-985539cc51 has been pushed to the Fedora 44 testing repository.
Soon you'll be able to install the update with the following command:
`sudo dnf upgrade --enablerepo=updates-testing --refresh --advisory=FEDORA-2026-985539cc51`
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2026-985539cc51

See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.

Comment 19 Kamil Páral 2026-04-20 10:42:19 UTC
Works fine in RC1.4. The drive is visible, I can erase the whole drive, or I can remove just the btrfs partition.

When removing the btrfs partition, there's a broken message, which is a bit ugly, but it doesn't prevent the removal.

Comment 20 Kamil Páral 2026-04-20 10:43:04 UTC
Created attachment 2137682 [details]
broken message during partition removal

This is how partition removal looks like in this case.

Comment 21 Fedora Update System 2026-04-24 05:56:18 UTC
FEDORA-2026-985539cc51 (python-blivet-3.13.2-2.fc44) has been pushed to the Fedora 44 stable repository.
If problem still persists, please make note of it in this bug report.


Note You need to log in before you can comment on or make changes to this bug.