fix(orion): retain Antares mounts and improve dicfuse startup diagnostics (#2026)

Signed-off-by: Luxian <lux1an@qq.com>

Author: Ivanbeethoven

orion/src/antares.rs

@@ -3,12 +3,19 @@
 //! This module provides a singleton wrapper around `scorpiofs::AntaresManager`
 //! for managing overlay filesystem mounts used during build operations.
 
-use std::{error::Error, io, path::PathBuf, sync::Arc, time::Duration};
+use std::{
+    error::Error,
+    io,
+    path::{Path, PathBuf},
+    sync::Arc,
+    time::Duration,
+};
 
 use scorpiofs::{AntaresConfig, AntaresManager, AntaresPaths};
 use tokio::sync::OnceCell;
 
 static MANAGER: OnceCell<Arc<AntaresManager>> = OnceCell::const_new();
+const TEST_BROWSE_JOB_ID: &str = "antares_test";
 
 type DynError = Box<dyn Error + Send + Sync>;
 
@@ -128,6 +135,7 @@ pub async fn mount_job(job_id: &str, cl: Option<&str>) -> Result<AntaresConfig,
 pub(crate) async fn warmup_dicfuse() -> Result<(), DynError> {
     tracing::info!("Initializing Antares Dicfuse during Orion startup");
     let manager = get_manager().await?;
+    let manager_for_test_mount = Arc::clone(manager);
     let dicfuse = manager.dicfuse();
 
     // Idempotent: safe even if the manager already started import internally.
@@ -149,7 +157,17 @@ pub(crate) async fn warmup_dicfuse() -> Result<(), DynError> {
         )
         .await
         {
-            Ok(_) => tracing::info!("Antares Dicfuse warmup completed"),
+            Ok(_) => {
+                tracing::info!("Antares Dicfuse warmup completed");
+                log_dicfuse_root_tree();
+                if is_test_mount_enabled() {
+                    ensure_test_mount(manager_for_test_mount.as_ref()).await;
+                } else {
+                    tracing::info!(
+                        "Antares test mount disabled by ORION_ENABLE_ANTARES_TEST_MOUNT"
+                    );
+                }
+            }
             Err(_) => tracing::warn!(
                 "Antares Dicfuse warmup timed out after {}s",
                 warmup_timeout_secs
@@ -160,6 +178,131 @@ pub(crate) async fn warmup_dicfuse() -> Result<(), DynError> {
     Ok(())
 }
 
+fn is_test_mount_enabled() -> bool {
+    match std::env::var("ORION_ENABLE_ANTARES_TEST_MOUNT") {
+        Ok(v) => {
+            let v = v.trim().to_ascii_lowercase();
+            !(v == "0" || v == "false" || v == "no" || v == "off")
+        }
+        Err(_) => true,
+    }
+}
+
+async fn ensure_test_mount(manager: &AntaresManager) {
+    match manager.mount_job(TEST_BROWSE_JOB_ID, None).await {
+        Ok(config) => {
+            tracing::info!(
+                "Antares test mount ready: job_id={}, mountpoint={}",
+                TEST_BROWSE_JOB_ID,
+                config.mountpoint.display()
+            );
+        }
+        Err(err) => {
+            tracing::warn!(
+                "Failed to create Antares test mount job_id={}: {}",
+                TEST_BROWSE_JOB_ID,
+                err
+            );
+        }
+    }
+}
+
+fn log_dicfuse_root_tree() {
+    let root = PathBuf::from(scorpiofs::util::config::workspace());
+    let max_depth = std::env::var("ORION_DICFUSE_ROOT_TREE_DEPTH")
+        .ok()
+        .and_then(|v| v.parse::<usize>().ok())
+        .unwrap_or(2);
+    let max_entries = std::env::var("ORION_DICFUSE_ROOT_TREE_MAX_ENTRIES")
+        .ok()
+        .and_then(|v| v.parse::<usize>().ok())
+        .unwrap_or(200);
+
+    tracing::info!(
+        root = %root.display(),
+        max_depth,
+        max_entries,
+        "Dicfuse init: printing workspace root tree"
+    );
+
+    if !root.exists() {
+        tracing::warn!("Dicfuse workspace path does not exist: {}", root.display());
+        return;
+    }
+
+    let mut printed = 0usize;
+    tracing::info!("[dicfuse-root] /");
+    log_tree_recursive(&root, &root, 0, max_depth, max_entries, &mut printed);
+
+    if printed >= max_entries {
+        tracing::info!(
+            "Dicfuse root tree output truncated at {} entries (set ORION_DICFUSE_ROOT_TREE_MAX_ENTRIES to increase)",
+            max_entries
+        );
+    }
+}
+
+fn log_tree_recursive(
+    root: &Path,
+    current: &Path,
+    depth: usize,
+    max_depth: usize,
+    max_entries: usize,
+    printed: &mut usize,
+) {
+    if depth >= max_depth || *printed >= max_entries {
+        return;
+    }
+
+    let entries = match std::fs::read_dir(current) {
+        Ok(entries) => entries,
+        Err(err) => {
+            tracing::warn!("Failed to read {}: {}", current.display(), err);
+            return;
+        }
+    };
+
+    let mut children: Vec<(String, PathBuf, bool)> = Vec::new();
+    for entry in entries {
+        let entry = match entry {
+            Ok(entry) => entry,
+            Err(err) => {
+                tracing::warn!("read_dir entry error under {}: {}", current.display(), err);
+                continue;
+            }
+        };
+
+        let path = entry.path();
+        let name = entry.file_name().to_string_lossy().to_string();
+        let is_dir = entry.file_type().map(|t| t.is_dir()).unwrap_or(false);
+        children.push((name, path, is_dir));
+    }
+
+    children.sort_by(|a, b| a.0.cmp(&b.0));
+
+    for (_name, path, is_dir) in children {
+        if *printed >= max_entries {
+            return;
+        }
+
+        let rel = path
+            .strip_prefix(root)
+            .map(|p| p.display().to_string())
+            .unwrap_or_else(|_| path.display().to_string());
+        let indent = "  ".repeat(depth + 1);
+        if is_dir {
+            tracing::info!("[dicfuse-root] {}{}/", indent, rel);
+        } else {
+            tracing::info!("[dicfuse-root] {}{}", indent, rel);
+        }
+        *printed += 1;
+
+        if is_dir {
+            log_tree_recursive(root, &path, depth + 1, max_depth, max_entries, printed);
+        }
+    }
+}
+
 /// Unmount and cleanup a job overlay filesystem.
 ///
 /// # Arguments

orion/src/buck_controller.rs

@@ -4,7 +4,6 @@ use std::{
     io::BufReader,
     path::{Path, PathBuf},
     process::{ExitStatus, Stdio},
-    sync::atomic::{AtomicBool, Ordering},
 };
 
 use anyhow::anyhow;
@@ -46,14 +45,6 @@ static PROJECT_ROOT: Lazy<String> =
 const DEFAULT_PREHEAT_SHALLOW_DEPTH: usize = 3;
 static BUILD_CONFIG: Lazy<Option<BuildConfig>> = Lazy::new(load_build_config);
 
-/// Check whether failed-build mounts should be kept alive for debugging.
-/// Controlled by the `ORION_KEEP_FAILED_MOUNTS` environment variable (set to "1" to enable).
-fn keep_failed_mounts() -> bool {
-    std::env::var("ORION_KEEP_FAILED_MOUNTS")
-        .map(|v| v == "1")
-        .unwrap_or(false)
-}
-
 /// Mount an Antares overlay filesystem for a build job.
 ///
 /// Creates a new Antares overlay mount using scorpiofs. The underlying Dicfuse
@@ -512,63 +503,6 @@ async fn flush_buffer(
     }
 }
 
-/// RAII guard for automatically unmounting Antares filesystem when dropped
-struct MountGuard {
-    mount_id: String,
-    task_id: String,
-    unmounted: AtomicBool,
-}
-
-impl MountGuard {
-    fn new(mount_id: String, task_id: String) -> Self {
-        Self {
-            mount_id,
-            task_id,
-            unmounted: AtomicBool::new(false),
-        }
-    }
-
-    async fn unmount(&self) {
-        if self.unmounted.swap(true, Ordering::AcqRel) {
-            return;
-        }
-        match unmount_antares_fs(&self.mount_id).await {
-            Ok(_) => tracing::info!("[Task {}] Filesystem unmounted successfully.", self.task_id),
-            Err(e) => {
-                tracing::error!(
-                    "[Task {}] Failed to unmount filesystem: {}",
-                    self.task_id,
-                    e
-                )
-            }
-        }
-    }
-}
-
-impl Drop for MountGuard {
-    fn drop(&mut self) {
-        if self.unmounted.load(Ordering::Acquire) {
-            return;
-        }
-        if keep_failed_mounts() {
-            tracing::info!(
-                "[Task {}] MountGuard dropped — unmount skipped (ORION_KEEP_FAILED_MOUNTS=1, mount_id={}).",
-                self.task_id,
-                self.mount_id,
-            );
-            return;
-        }
-        let mount_id = self.mount_id.clone();
-        let task_id: String = self.task_id.clone();
-        tokio::spawn(async move {
-            match unmount_antares_fs(&mount_id).await {
-                Ok(_) => tracing::info!("[Task {}] Filesystem unmounted successfully.", task_id),
-                Err(e) => tracing::error!("[Task {}] Failed to unmount filesystem: {}", task_id, e),
-            }
-        });
-    }
-}
-
 /// Executes buck build with filesystem mounting and output streaming.
 ///
 /// Process flow:
@@ -612,8 +546,6 @@ pub async fn build(
     const MAX_TARGETS_ATTEMPTS: usize = 2;
     let mut mount_point = None;
     let mut old_repo_mount_point_saved = None;
-    let mut mount_guard = None;
-    let mut mount_guard_old_repo = None;
     let mut targets: Vec<TargetLabel> = Vec::new();
     let mut last_targets_error: Option<anyhow::Error> = None;
 
@@ -629,13 +561,11 @@ pub async fn build(
         // Buck2 isolates daemons by project root, so distinct mount paths
         // naturally get separate daemons without needing `--isolation-dir`.
         let id_for_old_repo = format!("{id}-old-{attempt}");
-        let (old_repo_mount_point, mount_id_old_repo) =
+        let (old_repo_mount_point, _mount_id_old_repo) =
             mount_antares_fs(&id_for_old_repo, None).await?;
-        let guard_old_repo = MountGuard::new(mount_id_old_repo.clone(), id_for_old_repo);
 
         let id_for_repo = format!("{id}-{attempt}");
-        let (repo_mount_point, mount_id) = mount_antares_fs(&id_for_repo, cl_arg).await?;
-        let guard = MountGuard::new(mount_id.clone(), id_for_repo);
+        let (repo_mount_point, _mount_id) = mount_antares_fs(&id_for_repo, cl_arg).await?;
 
         tracing::info!(
             "[Task {}] Filesystem mounted successfully (attempt {}/{}).",
@@ -658,23 +588,12 @@ pub async fn build(
             Ok(found_targets) => {
                 mount_point = Some(repo_mount_point);
                 old_repo_mount_point_saved = Some(old_repo_mount_point.clone());
-                mount_guard = Some(guard);
-                mount_guard_old_repo = Some(guard_old_repo);
                 targets = found_targets;
                 break;
             }
             Err(e) => {
-                if keep_failed_mounts() {
-                    tracing::info!(
-                        "[Task {}] Keeping failed mounts alive for debugging (ORION_KEEP_FAILED_MOUNTS=1)",
-                        id,
-                    );
-                } else {
-                    guard.unmount().await;
-                    guard_old_repo.unmount().await;
-                }
                 tracing::warn!(
-                    "[Task {}] Failed to get build targets (attempt {}/{}): {}. Mounts kept alive for debugging (old={}, new={}).",
+                    "[Task {}] Failed to get build targets (attempt {}/{}): {}. Mounts retained for debugging (old={}, new={}).",
                     id,
                     attempt,
                     MAX_TARGETS_ATTEMPTS,
@@ -712,9 +631,6 @@ pub async fn build(
             return Err(err.into());
         }
     };
-    let mount_guard = mount_guard.ok_or("Mount guard missing after target discovery")?;
-    let mount_guard_old_repo =
-        mount_guard_old_repo.ok_or("Old repo mount guard missing after target discovery")?;
 
     let build_result = async {
         // Run buck2 build from the sub-project directory, not the monorepo root.
@@ -830,41 +746,14 @@ pub async fn build(
     }
     .await;
 
-    if keep_failed_mounts() {
-        tracing::info!(
-            "[Task {}] Skipping unmount (ORION_KEEP_FAILED_MOUNTS=1) — mount directories retained: \
-             new_repo mountpoint={}, mount_id={}; \
-             old_repo mountpoint={}, mount_id={}",
-            id,
-            mount_point,
-            mount_guard.mount_id,
-            old_repo_mount_point_saved.as_deref().unwrap_or("<unknown>"),
-            mount_guard_old_repo.mount_id,
-        );
-        // Prevent the Drop impl from unmounting.
-        mount_guard.unmounted.store(true, Ordering::Release);
-        mount_guard_old_repo
-            .unmounted
-            .store(true, Ordering::Release);
-    } else {
-        mount_guard.unmount().await;
-        mount_guard_old_repo.unmount().await;
-    }
+    tracing::info!(
+        "[Task {}] Build completed — mount directories retained for debugging: \
+         new_repo mountpoint={}; \
+         old_repo mountpoint={}",
+        id,
+        mount_point,
+        old_repo_mount_point_saved.as_deref().unwrap_or("<unknown>"),
+    );
 
     build_result
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[tokio::test]
-    async fn test_mount_guard_creation() {
-        let mount_guard = MountGuard::new("test_mount_id".to_string(), "test_task_id".to_string());
-        assert_eq!(mount_guard.mount_id, "test_mount_id");
-        assert_eq!(mount_guard.task_id, "test_task_id");
-    }
-
-    // Note: mount/unmount tests removed - they now use scorpiofs direct calls
-    // which require actual filesystem setup. See integration tests instead.
-}
+}

orion/systemd/orion-runner.service

@@ -30,7 +30,8 @@ CapabilityBoundingSet=CAP_SYS_ADMIN CAP_DAC_OVERRIDE CAP_DAC_READ_SEARCH
 
 # Basic hardening (relaxed for FUSE operations)
 NoNewPrivileges=false
-PrivateTmp=true
+PrivateTmp=false
+PrivateMounts=no
 
 [Install]
 WantedBy=multi-user.target