fix(windows): strip \\?\ prefix from canonicalized paths before disk mount-point matching (#13045)

Fixes #13041.

## Root cause

Two broken string comparisons, both caused by Windows path
canonicalization:

**1. `src/config/src/config.rs` — `check_disk_cache_config()`**

`Path::canonicalize()` on Windows returns a verbatim extended-length
path (`\\?\C:\OpenObserve\data\cache`), but `sysinfo` returns plain
mount points (`C:\`). The `starts_with` check
`"\\?\C:\...".starts_with("C:\\")` is always `false` → `disk_total =
disk_free = 0` on every Windows deployment.

**2. `src/job/stats.rs` — `update_node_disk_usage()`**

The comparison was also reversed: `mount_point.starts_with(data_dir)`
checks whether the mount point path begins with the (much longer) data
directory path — `"C:\\".starts_with("C:\\OpenObserve\\data\\")` is
always `false`. The correct direction is
`data_dir.starts_with(mount_point)`.

## Consequences when disk sizes are 0

- Disk cache `max_size` is computed from near-zero `effective_free`,
leaving `result_max_size = 0`
- `disk.rs` GC panics at first tick: `max_size / result_max_size`
evaluates before `max(1, …)` → divide by zero, GC task dies permanently
- `check_disk_circuit_breaker()` always compares `0 > 0` → never trips →
no self-protection against disk exhaustion on Windows

## Fix

1. **`config.rs`**: strip the `\\?\` prefix from the canonicalized cache
dir string before the `starts_with` lookup against sysinfo mount points.
2. **`stats.rs`**: fix the reversed comparison
(`mount_point.starts_with(data_dir)` →
`data_dir.starts_with(mount_point)`) and apply the same `\\?\` prefix
strip so canonicalized data dirs match correctly.

Both fixes are no-ops on Linux/macOS where `canonicalize()` never
produces the `\\?\` prefix.

---------

Co-authored-by: Hengfei Yang <hengfei.yang@gmail.com>

src/config/src/config.rs

@@ -3252,12 +3252,36 @@ fn check_memory_config(cfg: &mut Config) -> Result<(), anyhow::Error> {
     Ok(())
 }
 
+/// Strip the Windows extended-length prefix (`\\?\`) from a canonicalized path
+/// so it can be compared with sysinfo mount points that use the plain DOS form.
+///
+/// Uses [`std::path::Prefix`] to detect verbatim prefixes rather than
+/// manipulating the string directly, which would silently break on non-ASCII
+/// drive letters or UNC paths.
+pub fn deverbatim(path: &Path) -> std::borrow::Cow<'_, str> {
+    #[cfg(windows)]
+    {
+        use std::path::{Component, Prefix};
+        if let Some(Component::Prefix(p)) = path.components().next() {
+            if let Prefix::VerbatimDisk(drive) = p.kind() {
+                // \\?\C:\rest → C:\rest
+                // p.as_os_str() is "\\?\C:" (6 bytes); the remainder of the
+                // original string is "\rest", so prepend the plain drive letter.
+                let after_prefix = &path.to_string_lossy()[p.as_os_str().len()..];
+                return format!("{}:{}", drive as char, after_prefix).into();
+            }
+        }
+    }
+    path.to_string_lossy()
+}
+
 fn check_disk_cache_config(cfg: &mut Config) -> Result<(), anyhow::Error> {
     std::fs::create_dir_all(&cfg.common.data_cache_dir).expect("create cache dir success");
-    let cache_dir = Path::new(&cfg.common.data_cache_dir)
+    let cache_dir_path = Path::new(&cfg.common.data_cache_dir)
         .canonicalize()
         .unwrap();
-    let cache_dir = cache_dir.to_str().unwrap();
+    let cache_dir_owned = deverbatim(&cache_dir_path).into_owned();
+    let cache_dir = cache_dir_owned.as_str();
 
     // disable disk cache for local disk storage
     if cfg.common.is_local_storage
@@ -4333,4 +4357,34 @@ mod tests {
         let name = get_cluster_name();
         assert!(!name.is_empty(), "cluster name should not be empty");
     }
+
+    #[test]
+    fn test_deverbatim_plain_path_unchanged() {
+        let p = std::path::Path::new("/data/openobserve");
+        let result = deverbatim(p);
+        assert_eq!(result, "/data/openobserve");
+    }
+
+    #[test]
+    fn test_deverbatim_empty_path_unchanged() {
+        let p = std::path::Path::new("");
+        let result = deverbatim(p);
+        assert_eq!(result, "");
+    }
+
+    #[cfg(windows)]
+    #[test]
+    fn test_deverbatim_verbatim_disk_stripped() {
+        let p = std::path::Path::new(r"\\?\C:\data\openobserve");
+        let result = deverbatim(p);
+        assert_eq!(result, r"C:\data\openobserve");
+    }
+
+    #[cfg(windows)]
+    #[test]
+    fn test_deverbatim_plain_windows_path_unchanged() {
+        let p = std::path::Path::new(r"C:\data\openobserve");
+        let result = deverbatim(p);
+        assert_eq!(result, r"C:\data\openobserve");
+    }
 }

src/job/stats.rs

@@ -13,7 +13,7 @@
 // You should have received a copy of the GNU Affero General Public License
 // along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
-use config::{cluster::LOCAL_NODE, get_config, spawn_pausable_job};
+use config::{cluster::LOCAL_NODE, deverbatim, get_config, spawn_pausable_job};
 use tokio::time;
 
 use crate::service::{compact::stats::update_stats_from_file_list, db};
@@ -112,21 +112,22 @@ async fn update_node_memory_usage() -> Result<(), anyhow::Error> {
 // update node disk usage metrics every 60 seconds
 async fn update_node_disk_usage() -> Result<(), anyhow::Error> {
     let cfg = get_config();
-    let data_dir = std::path::Path::new(&cfg.common.data_dir);
+    let data_dir_raw = std::path::Path::new(&cfg.common.data_dir)
+        .canonicalize()
+        .unwrap_or_else(|_| std::path::PathBuf::from(&cfg.common.data_dir));
+    // Strip any Windows extended-length prefix (\\?\) returned by canonicalize
+    // so comparisons with sysinfo mount points (plain DOS form) work correctly.
+    let data_dir_norm = deverbatim(&data_dir_raw).into_owned();
 
     loop {
         let disks = config::utils::sysinfo::disk::get_disk_usage();
-        // Sum up all disks that contain subdirectories of the data directory
-        // This handles cases where multiple disks are mounted at different subpaths
+        // Sum up all disks whose mount point is a prefix of the data directory.
+        // This finds the disk(s) that actually host the data directory.
         let mut total_space = 0_u64;
         let mut total_used = 0_u64;
 
         for disk in disks {
-            // Check if the disk mount point is under the data directory
-            if disk
-                .mount_point
-                .starts_with(data_dir.to_string_lossy().as_ref())
-            {
+            if data_dir_norm.starts_with(disk.mount_point.as_str()) {
                 total_space += disk.total_space;
                 total_used += disk.total_space - disk.available_space;
             }