# A couple of minor issues on first boot

**URL:** <https://universal-blue.discourse.group/t/a-couple-of-minor-issues-on-first-boot/6040>\
**Category:** Bluefin\
**Created:** [January 1, 2025, 6:31pm UTC](https://universal-blue.discourse.group/t/a-couple-of-minor-issues-on-first-boot/6040 "2025-01-01T18:31:39Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![arora9090](https://avatars.discourse-cdn.com/v4/letter/a/45deac/32.png) [@arora9090](https://universal-blue.discourse.group/u/arora9090)\
**Post date:** [January 1, 2025, 6:31pm UTC](https://universal-blue.discourse.group/t/a-couple-of-minor-issues-on-first-boot/6040/1 "2025-01-01T18:31:39Z")

</div>

First of all, thank you very much for a very stable and sane distro! I have been running Aurora as my primary workstation for last year or so and recently installed it on other machines at my home.

On that note, I noticed a few minor issues on first boot none of which are show stoppers, but I thought it might be good to point them out in case someone else runs into them.

1. `/var/lib/setroubleshoot` directory is missing which leads to the `setroubleshootd` service not starting properly. Had to do the below to fix…

```sh
sudo mkdir /var/lib/setroubleshoot
sudo chown setroubleshoot: /var/lib/setroubleshoot
sudo restorecon -R -v /var/lib/setroubleshoot
sudo systemctl restart setroubleshootd

```

Also, it might be worth installing the `setroubleshoot` package, which is a nice GUI tool to debug and troubleshoot selinux issues, in the base images. It is not available as a flatpak, so had to install it via rpm-ostree install.

1. The userid and groupid are hardcodded to 1000 in `/usr/lib/systemd/system/brew-<setup|update|upgrade>` systemd service files. This is fine if the default user is created as is, but a problem if the user is created with a diff. uid/gid than 1000. I have a NFS server at home, and hence need the default uid/gid to be diff. that 1000. I had to fix this using below…

```sh
# From the logged in user
sudo chown -R $USER: /home/linuxbrew
sudo cp /usr/lib/systemd/system/brew-{setup,update,upgrade} /etc/systemd/system/.
# Change 1000 to the desired uid/gid in the /etc/systemd/system/brew-* scripts

```

1. hostname if set during the anaconda installer is not persisted and overwritten by the transient hostname. Had to use `sudo hostnamectl hostname <DESIRED HOSTNAME>` to set after first boot.

---

<div class="post-metadata">

**Author:** ![akvadrako](https://yyz2.discourse-cdn.com/free1/user_avatar/universal-blue.discourse.group/akvadrako/32/4150_2.png) [@akvadrako](https://universal-blue.discourse.group/u/akvadrako)\
**Post date:** [February 21, 2025, 11:59am UTC](https://universal-blue.discourse.group/t/a-couple-of-minor-issues-on-first-boot/6040/2 "2025-02-21T11:59:13Z")

</div>

Related Github issue for the missing setroubleshoot directories.

> <https://github.com/coreos/fedora-coreos-tracker/issues/1720>
>
> \### Describe the bug
> 
> Cockpit is not showing selinux events even though they are… available via: \`journbalctl -t setroubleshoot\`
> 
> After debuging the problem, I noticed that the directory \`/var/lib/setroubleshoot\` is not created when layering \`troubleshoot-server\` package. As such, applications depending on \`/var/lib/setroubleshoot/setroubleshoot\_database.xml\` like \`cockpit-selinux\` are blind.
> 
> \*\*Note:\*\*
> The same happens in Silverblue, and SELinux Troubleshoot (UI) doesn't show the events.
> 
> 
> 
> \### Reproduction steps
> 
> \*\*Prerequisites:\*\*
> \`cockpit\` and \`cockpit-selinux\` needs to be installed. 
> cockpit needs to be running.
> 
> 1. \`rpm-ostree install setroubleshoot-server\`
> 2. Reboot
> 3. Trigger a selinux event: \`podman run --detach --name mariadb --env MARIADB\_ALLOW\_EMPTY\_ROOT\_PASSWORD=1 mariadb:latest\`. This container requires \`container\_manage\_cgroup\` boolean.
> 4. Confirm that selinux triggered an event: 
> \`journalctl -t setroubleshoot\` OR
> \`ausearch -m AVC\`
> 5. Login to Cockpit and open SELinux tab. It won't show any event.
> 6. Review the messages from setroubleshootd:
> \`journalctl -u setroubleshootd\` It will show errors related to not being able to create \`/var/lib/setroubleshoot/setroubleshoot\_database.xml\`
> 
> \### Expected behavior
> 
> All selinux events shoud be accessible from cockpit and SELinux Troubleshoot (UI) in Silverblue.
> 
> \### Actual behavior
> 
> Events not available by third party apps.
> 
> \### System details
> 
> Fedora CoreOS - Vultr
> 
> 
> \### Butane or Ignition config
> 
> \_No response\_
> 
> \### Additional information
> 
> The problem can be resolved by manually creating the missing directory:
> 
> \`\`\`
> mkdir /var/lib/setroubleshoot
> chown setroubleshoot:setroubleshoot /var/lib/setroubleshoot
> chmod 0700 /var/lib/setroubleshoot
> \`\`\`

As I understand, on ostree systems /var is never touched by RPMs. So they need correct tmpfiles.d configs and setroubleshootd doesn’t have one.

---

<div class="post-metadata">

**Author:** ![system](https://global.discourse-cdn.com/free1/uploads/univeral_blue/original/2X/e/e39f7d1952979cf23b0f8d634e025fb38da50026.png) [@system](https://universal-blue.discourse.group/u/system)\
**Post date:** [May 22, 2025, 11:59am UTC](https://universal-blue.discourse.group/t/a-couple-of-minor-issues-on-first-boot/6040/3 "2025-05-22T11:59:42Z")

</div>

Closed on May 22, 2025, 11:59am UTC

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
