Fedora 29 Silverblue Stable - Fatal error

Project Discussion
1

Hi folks,
I’m currently experiencing issues attempting to install the stable release of Fedora 29 Silverblue (as are others). The error itself is listed below and seems to be relating to grub installation during the Anaconda install process.

The following error ocurred during the installation. This is fatal and the installation will be aborted.

ostree [‘admin’, ‘-sysroot=/mnt/sysimage’,‘deploy’,’–os=fedora-workstation’, ‘fedora-workstation:fedora/29/x86_64/silverblue’’ exited with code -6

Reproduction

  • Boot Fedora 29 Silverblue from USB drive
  • Follow installation process - mapping the /boot/efi partition to /dev/sda1 (EFI)

Expected result
Installation completes successfully.

Actual result
Installation fails.

Additional info
I suspect this is somehow related to existing installations being present.

GPT
/dev/sda1 - EFI System, FAT 32   /boot/efi
/dev/sda2 - APM
/dev/sda3 - EXT4
/dev/sda4 - EXT4
/dev/sda5 - NTFS
/dev/sda6 - EXT4                 /

If anyone has any thoughts or suggestions as to what may be causing this, I would hugely appreciate some help. There are a number of existing discussions relating this issue both on here and Bugzilla.

Other references

2

Can you try following these steps to access the program log and see what it shows?

1 Like
3

Thanks for the response, logs are below.

Errors

program.log

16:33:07,700 INF program: Running... ostree admin --sysroot=/mnt/sysimage deploy --os=fedora-workstation fedora-workstation:fedora/29/x86_64/silverblue
16:33:13,320 INF program: Relabeling /var (no stamp file 'var/.ostree-selabeled' found)
16:33:13,321 INF program: **
16:33:13,321 INF program: OSTree:ERROR:src/libostree/ostree-bootloader-grub2.c:354:_ostree_bootloader_grub2_write_config: assertion failed (deployments->len > 0): (0 > 0)
16:33:13,321 DBG program: Return code: -6

storage.log

16:30:57,870 DBG blivet: get_format('efivarfs') returning EFIVarFS instance with object id 604
16:30:57,870 DBG blivet: get_format('None') returning DeviceFormat instance with object id 606
16:30:57,870 DBG blivet:                       NoDevice._set_format: efivarfs ; type: efivarfs ; current: None ;
16:30:57,871 DBG blivet:                     NoDevice.update_sysfs_path: efivarfs ; status: False ;
16:30:57,871 ERR blivet: failed to update sysfs path for efivarfs: [Errno 2] No such file or directory: 'efivarfs'
16:30:57,872 DBG blivet:                   LVMLogicalVolumeDevice.setup: fedora00-root ; orig: False ; status: True ; controllable: True ;
16:30:57,897 DBG blivet:                Ext4FS.setup: device: /dev/mapper/fedora00-root ; type: ext4 ; status: False ;
16:30:57,930 DBG blivet:                PartitionDevice.setup: sda1 ; orig: False ; status: True ; controllable: True ;
16:30:58,011 DBG blivet:                EFIFS.setup: device: /dev/sda1 ; type: efi ; status: False ;
16:30:58,041 INF blivet: failed to set SELinux context for /mnt/sysimage/boot/efi: [Errno 95] Operation not supported
16:30:58,041 WRN blivet: Failed to reset SElinux context for newly mounted filesystem root directory to default.
16:30:58,041 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /boot/efi/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,042 DBG blivet:                DirectoryDevice.setup: /dev ; orig: False ; status: True ; controllable: True ;
16:30:58,164 DBG blivet:                BindFS.setup: device: /dev ; type: bind ; status: False ;
16:30:58,192 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /dev/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,193 DBG blivet:                NoDevice.setup: devpts ; orig: False ; status: False ; controllable: True ;
16:30:58,294 DBG blivet:                DevPtsFS.setup: device: devpts ; type: devpts ; status: False ;
16:30:58,319 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /dev/pts/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,320 DBG blivet:                NoDevice.setup: tmpfs ; orig: False ; status: False ; controllable: True ;
16:30:58,399 DBG blivet:                TmpFS.setup: device: tmpfs ; type: tmpfs ; status: True ;
16:30:58,406 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /dev/shm/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,407 DBG blivet:                NoDevice.setup: proc ; orig: False ; status: False ; controllable: True ;
16:30:58,512 DBG blivet:                ProcFS.setup: device: proc ; type: proc ; status: False ;
16:30:58,542 INF blivet: failed to set SELinux context for /mnt/sysimage/proc: [Errno 95] Operation not supported
16:30:58,543 WRN blivet: Failed to reset SElinux context for newly mounted filesystem root directory to default.
16:30:58,543 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /proc/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,578 DBG blivet:                DirectoryDevice.setup: /run ; orig: False ; status: True ; controllable: True ;
16:30:58,695 DBG blivet:                BindFS.setup: device: /run ; type: bind ; status: False ;
16:30:58,723 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /run/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,724 DBG blivet:                NoDevice.setup: sysfs ; orig: False ; status: False ; controllable: True ;
16:30:58,828 DBG blivet:                SysFS.setup: device: sysfs ; type: sysfs ; status: False ;
16:30:58,854 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /sys/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,855 DBG blivet:                NoDevice.setup: efivarfs ; orig: False ; status: False ; controllable: True ;
16:30:58,956 DBG blivet:                EFIVarFS.setup: device: efivarfs ; type: efivarfs ; status: False ;
16:30:58,981 INF blivet: failed to set SELinux context for /mnt/sysimage/sys/firmware/efi/efivars: [Errno 95] Operation not supported
16:30:58,981 WRN blivet: Failed to reset SElinux context for newly mounted filesystem root directory to default.
16:30:58,981 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /sys/firmware/efi/efivars/lost+found to system_u:object_r:lost_found_t:s0
16:30:58,982 DBG blivet:                NoDevice.setup: selinuxfs ; orig: False ; status: False ; controllable: True ;
16:30:59,080 DBG blivet:                SELinuxFS.setup: device: selinuxfs ; type: selinuxfs ; status: False ;
16:30:59,109 INF blivet: failed to set SELinux context for /mnt/sysimage/sys/fs/selinux: [Errno 95] Operation not supported
16:30:59,109 WRN blivet: Failed to reset SElinux context for newly mounted filesystem root directory to default.
16:30:59,109 WRN blivet: Failed to set SELinux context for newly mounted filesystem lost+found directory at /sys/fs/selinux/lost+found to system_u:object_r:lost_found_t:s0

anaconda.log

16:33:07,381 INF payload.rpmostreepayload: ostree pull: 6790 metadata, 92403 content objects imported
16:33:07,700 INF payload.rpmostreepayload: ostree admin deploy starting

Full Logs

Google drive

4

Are you using automatic partitioning or manual?

5

Manual. I’ve tried every possible combination I could think of:

  • With/without LVM
  • With/without LUKS
  • With/without separate /boot

Fedora 29 (non Silverblue) installs without an issue.

6

What is the exact layout you’re using? Have you tried going in to manual partitioning but having it auto-create the ones you need, then adjusting them as necessary? In general, the Silverblue installer doesn’t yet cater to anything that’s different from the default, especially requiring a separate /boot and /boot/efi (especially if you forgot /boot/efi).

7

Hey, apologies for not getting back to this sooner.

It’s definitely not a partition layout issue. I’ve managed to do clean installs with identical partition layouts using Fedora 29 (non silverblue).

An example of one of the many partition layouts I’ve tried is in the original post :slight_smile:

8

Silverblue uses the same installer base as Fedora, but the installation steps taken themselves are quite different, and if anything is different than it expects, it tends to explode a bit.

Maybe try creating a new /boot/efi and then merging them manually post-installation? (Multiple ESPs are supported perfectly.) You should be able to just copy all the .efi files over back to the original ESP and then change it in /etc/fstab.

9

@refi64 @andyrichardson I’ve finally found the perfect post. I had yesterday identically the same issue with the same error ! No issue observed with F29 (non silverblue) and no issue with fedora 30 silverblue installation on a virtual machine. I’m actually sentenced to use it through a vm :frowning:

Any evolution here?

Thanks.

10

@refi64 @andyrichardson Guys???

11

Was the error identical in the logs? The assertion failure?

12

@refi64: Below error was the same (during the install) but I didn’t checked the logs yet. I’ll check them ASAP and let you know.

admin’, ‘-sysroot=/mnt/sysimage’,‘deploy’,’–os=fedora-workstation’, ‘fedora-workstation:fedora/29/x86_64/silverblue’’ exited with code -6

13

@refi64: I’m back with some news ! I’ve deleted first the GRUB from the previous F29 workstation installation (via windows). Then, I’ve tried another F30 silverblue automatic installation and I didn’t get above fatal error but another error concerning GRUB creation : impossible to create GRUB so your system will not boot up ! But I could ignore this error and continue the installation but OS didn’t boot. I’ve tried another installation but finally with manual partitioning and the installation has been completed without any issue ! Tested & approved ! I’m happy but this issue has to be fixed because I didn’t had any issue with automatic partitioning on F29 workstation. Thanks for your input :wink: