Rollup merge of #141643 - marcoieni:codebuild-check, r=Kobzol

ci: verify that codebuild jobs use ghcr.io

Author: jhpratt

src/ci/citool/src/jobs/tests.rs

@@ -1,4 +1,8 @@
+use std::path::Path;
+
+use super::Job;
 use crate::jobs::{JobDatabase, load_job_db};
+use crate::{DOCKER_DIRECTORY, JOBS_YML_PATH, utils};
 
 #[test]
 fn lookup_job_pattern() {
@@ -62,3 +66,65 @@ fn check_pattern(db: &JobDatabase, pattern: &str, expected: &[&str]) {
 
     assert_eq!(jobs, expected);
 }
+
+/// Validate that CodeBuild jobs use Docker images from ghcr.io registry.
+/// This is needed because otherwise from CodeBuild we get rate limited by Docker Hub.
+fn validate_codebuild_image(job: &Job) -> anyhow::Result<()> {
+    let is_job_on_codebuild = job.codebuild.unwrap_or(false);
+    if !is_job_on_codebuild {
+        // Jobs in GitHub Actions don't get rate limited by Docker Hub.
+        return Ok(());
+    }
+
+    let image_name = job.image();
+    // we hardcode host-x86_64 here, because in codebuild we only run jobs for this architecture.
+    let dockerfile_path =
+        Path::new(DOCKER_DIRECTORY).join("host-x86_64").join(&image_name).join("Dockerfile");
+
+    if !dockerfile_path.exists() {
+        return Err(anyhow::anyhow!(
+            "Dockerfile not found for CodeBuild job '{}' at path: {}",
+            job.name,
+            dockerfile_path.display()
+        ));
+    }
+
+    let dockerfile_content = utils::read_to_string(&dockerfile_path)?;
+
+    // Check if all FROM statement uses ghcr.io registry
+    let has_ghcr_from = dockerfile_content
+        .lines()
+        .filter(|line| line.trim_start().to_lowercase().starts_with("from "))
+        .all(|line| line.contains("ghcr.io"));
+
+    if !has_ghcr_from {
+        return Err(anyhow::anyhow!(
+            "CodeBuild job '{}' must use ghcr.io registry in its Dockerfile FROM statement. \
+                Dockerfile path: {dockerfile_path:?}",
+            job.name,
+        ));
+    }
+
+    Ok(())
+}
+
+#[test]
+fn validate_jobs() {
+    let db = {
+        let default_jobs_file = Path::new(JOBS_YML_PATH);
+        let db_str = utils::read_to_string(default_jobs_file).unwrap();
+        load_job_db(&db_str).expect("Failed to load job database")
+    };
+
+    let all_jobs =
+        db.pr_jobs.iter().chain(db.try_jobs.iter()).chain(db.auto_jobs.iter()).collect::<Vec<_>>();
+
+    let errors: Vec<anyhow::Error> =
+        all_jobs.into_iter().filter_map(|job| validate_codebuild_image(job).err()).collect();
+
+    if !errors.is_empty() {
+        let error_messages =
+            errors.into_iter().map(|e| format!("- {e}")).collect::<Vec<_>>().join("\n");
+        panic!("Job validation failed:\n{error_messages}");
+    }
+}

src/ci/citool/src/main.rs

@@ -27,7 +27,7 @@ use crate::test_dashboard::generate_test_dashboard;
 use crate::utils::{load_env_var, output_details};
 
 const CI_DIRECTORY: &str = concat!(env!("CARGO_MANIFEST_DIR"), "/..");
-const DOCKER_DIRECTORY: &str = concat!(env!("CARGO_MANIFEST_DIR"), "/../docker");
+pub const DOCKER_DIRECTORY: &str = concat!(env!("CARGO_MANIFEST_DIR"), "/../docker");
 const JOBS_YML_PATH: &str = concat!(env!("CARGO_MANIFEST_DIR"), "/../github-actions/jobs.yml");
 
 struct GitHubContext {