system-reinstall-bootc: fallback image from /usr/lib/os-release by hone · Pull Request #2135 · bootc-dev/bootc

hone · 2026-04-09T16:52:53Z

Add a fallback option when arguments are specified for an image. It works in this order:

BOOTC_REINSTALL_CONFIG env var to a YAML file
--image flag
reads /etc/os-release for BOOTC_IMAGE
reads /usr/lib/os-release for BOOTC_IMAGE

Add a fallback option when arguments are specified for an image. It works in this order: 1. BOOTC_REINSTALL_CONFIG env var to a YAML file 2. `--image` flag 3. reads /etc/os-release for BOOTC_IMAGE 4. reads /usr/lib/os-release for BOOTC_IMAGE Fixes: bootc-dev#1300 Signed-off-by: Terence Lee <hone02@gmail.com>

gemini-code-assist

Code Review

This pull request introduces a fallback mechanism to retrieve the bootc image from os-release files when not specified via CLI or environment variables. The changes include a new parsing module, refactored argument handling, and updated CI scripts. Feedback suggests parsing CLI arguments earlier to support standard flags like --help, filtering empty strings during os-release lookups for better fallback reliability, and consolidating redundant CI steps.

crates/system-reinstall-bootc/src/main.rs

gemini-code-assist · 2026-04-09T16:57:06Z

tmt/plans/integration.fmf

+    order: 97
    script:
      - mkdir -p bootc && cp /var/share/test-artifacts/*.src.rpm bootc
      - cd bootc && rpm2cpio *.src.rpm | cpio -idmv && rm -f *-vendor.tar.zstd && zstd -d *.tar.zstd && tar -xvf *.tar -C . --strip-components=1 && ls -al
      - pwd && ls -al && cd bootc/hack && ./provision-packit.sh
    when: running_env != image_mode
+  - how: shell
+    order: 98
+    script:
+      - echo 'BOOTC_IMAGE=localhost/bootc' | tee -a /usr/lib/os-release
+      - pwd && ls -al && cd bootc/hack && ./provision-packit.sh
+    when: running_env != image_mode


These two shell steps are redundant and inefficient. Both steps perform the same setup and run provision-packit.sh, which includes a time-consuming podman build operation. Since provision-packit.sh is already designed to handle both the explicit argument and the os-release fallback, you should combine these into a single step to avoid rebuilding the image and repeating the setup in CI.

order: 97 script: - mkdir -p bootc && cp /var/share/test-artifacts/*.src.rpm bootc - cd bootc && rpm2cpio *.src.rpm | cpio -idmv && rm -f *-vendor.tar.zstd && zstd -d *.tar.zstd && tar -xvf *.tar -C . --strip-components=1 && ls -al - pwd && ls -al && cd bootc/hack && ./provision-packit.sh - echo 'BOOTC_IMAGE=localhost/bootc' | tee -a /usr/lib/os-release - ./provision-packit.sh when: running_env != image_mode

I thought it was easier to just build the image twice than try to do a cleanup step for testing the BOOTC_IMAGE path on reinstall.

If cloud-init is detected in the target image: 1. Check the host for existing root SSH keys at /root/.ssh/authorized_keys. 2. If host keys exist, automatically configure inheritance via /target/root/.ssh/authorized_keys. 3. If no host keys exist, proceed without manual interaction, trusting cloud-init in the new system to handle key provisioning. If cloud-init is not detected, maintain the existinginteractive behavior to prevent user lock-out. Ref: bootc-dev#1300 Signed-off-by: Terence Lee <hone02@gmail.com>

hone · 2026-04-09T21:12:40Z

/gemini review

gemini-code-assist

Code Review

This pull request introduces the ability to automatically detect the bootc image from os-release files (/etc/os-release or /usr/lib/os-release) if not provided via the CLI. It also adds logic to automatically inherit host SSH keys when cloud-init is detected in the target image. Key changes include a new os_release module for parsing, updates to the CLI argument handling, and integration test adjustments. Feedback highlights a critical path error where a container-internal path is used as a host path for SSH key inheritance, potential inefficiencies and failures in the updated test plan, and an opportunity to refactor the image resolution logic for better maintainability.

crates/system-reinstall-bootc/src/main.rs

tmt/plans/integration.fmf

crates/system-reinstall-bootc/src/main.rs

Signed-off-by: Terence Lee <hone02@gmail.com>

Remove duplicate cloud-init from hack/packages.txt Signed-off-by: Terence Lee <hone02@gmail.com>

cgwalters

Thanks so much for working on this!

cgwalters · 2026-04-12T16:48:27Z

crates/system-reinstall-bootc/src/os_release.rs

+            continue;
+        }
+
+        if let Some((key, value)) = line.split_once('=') {


Style nit, also could be let/else here and avoid further nesting

cgwalters · 2026-04-12T21:01:41Z

crates/system-reinstall-bootc/src/os_release.rs

+                if value.starts_with('"') && value.ends_with('"') && value.len() >= 2 {
+                    let unquoted = &value[1..value.len() - 1];
+                    let processed = unquoted
+                        .replace(r#"\""#, "\"")
+                        .replace(r#"\\"#, "\\")
+                        .replace(r#"\$"#, "$")
+                        .replace(r#"\`"#, "`");


I think we can use shlex to parse this, it's already in the depchain

cgwalters · 2026-04-12T21:03:15Z

crates/system-reinstall-bootc/src/main.rs

+                .iter()
+                .find_map(|path| {
+                    os_release::get_bootc_image_from_file(path)
+                        .ok()


Generally avoid "error swallowing" like this. There's two different cases:

The file doens't exist - this is normal, and we have https://docs.rs/cap-std-ext/latest/cap_std_ext/dirext/trait.CapStdExtDirExt.html#tymethod.open_optional for this reason

All other errors (permission denied, failure to parse the file, etc.). I think for this we'd want at least to use

bootc/crates/utils/src/result_ext.rs

Line 4 in b1fe5d6

fn log_err_or_else<F>(self, default: F) -> T

)

cgwalters · 2026-04-12T21:04:18Z

crates/system-reinstall-bootc/src/podman.rs

 }

+#[context("image_has_cloud_init")]
+pub(crate) fn image_has_cloud_init(image: &str) -> Result<bool> {


In bootc-dev/bcvk#245 we're proposing to standardize labels on images that have Ignition, it could make sense to do the same for cloud-init.

Separate effort from this PR.

cgwalters · 2026-04-12T21:05:56Z

crates/system-reinstall-bootc/src/main.rs

+    if podman::image_has_cloud_init(&opts.image)? {
+        let host_root_keys = std::path::Path::new("/root/.ssh/authorized_keys");
+        if host_root_keys.exists() {
+            println!("Detected cloud-init and host keys. Inheriting keys automatically.");


I think this change may make sense but it looks completely unrelated to a fallback image approach right?

This seems like a distinct PR?

github-actions bot added the area/system-reinstall-bootc Issues related to system-reinstall-botoc label Apr 9, 2026

hone marked this pull request as draft April 9, 2026 16:53

bootc-bot bot requested a review from jmarrero April 9, 2026 16:53

gemini-code-assist bot reviewed Apr 9, 2026

View reviewed changes

crates/system-reinstall-bootc/src/main.rs Outdated Show resolved Hide resolved

tmt/plans/integration.fmf Show resolved Hide resolved

crates/system-reinstall-bootc/src/main.rs Outdated Show resolved Hide resolved

address gemini-code-assist feedback

ee263a3

Signed-off-by: Terence Lee <hone02@gmail.com>

hone force-pushed the default-image-os-release branch from 5bb1c1e to ee263a3 Compare April 9, 2026 22:43

hone marked this pull request as ready for review April 10, 2026 06:06

hone added 2 commits April 10, 2026 11:48

fix make validate

dcd8be0

Signed-off-by: Terence Lee <hone02@gmail.com>

Use --allowerase on cloud-init for centos package drift

a21dcba

Remove duplicate cloud-init from hack/packages.txt Signed-off-by: Terence Lee <hone02@gmail.com>

hone force-pushed the default-image-os-release branch from a168777 to a21dcba Compare April 10, 2026 16:49

cgwalters requested changes Apr 12, 2026

View reviewed changes

Conversation

hone commented Apr 9, 2026

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

gemini-code-assist bot Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

hone Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

hone commented Apr 9, 2026

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cgwalters left a comment

Choose a reason for hiding this comment

Uh oh!

cgwalters Apr 12, 2026

Choose a reason for hiding this comment

Uh oh!

cgwalters Apr 12, 2026

Choose a reason for hiding this comment

Uh oh!

cgwalters Apr 12, 2026

Choose a reason for hiding this comment

Uh oh!

cgwalters Apr 12, 2026

Choose a reason for hiding this comment

Uh oh!

cgwalters Apr 12, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants