Keep reconnect health breadcrumbs

src/trace.rs

@@ -1,11 +1,13 @@
 use std::fs::{File, OpenOptions};
 use std::io::Write;
-use std::path::PathBuf;
+use std::path::{Path, PathBuf};
 use std::sync::{Mutex, OnceLock};
 use std::time::{SystemTime, UNIX_EPOCH};
 
 static TRACE_FILE: OnceLock<Option<Mutex<File>>> = OnceLock::new();
+static HEALTH_FILE: OnceLock<Option<Mutex<File>>> = OnceLock::new();
 static TRACE_BYTES: OnceLock<bool> = OnceLock::new();
+const HEALTH_LOG_MAX_BYTES: u64 = 1024 * 1024;
 
 pub fn enabled() -> bool {
     TRACE_FILE.get_or_init(open_trace_file).is_some()
@@ -19,6 +21,21 @@ pub fn event(name: &str, fields: &[(&str, String)]) {
     let Some(file) = TRACE_FILE.get_or_init(open_trace_file) else {
         return;
     };
+    write_event(file, name, fields);
+}
+
+pub fn health_event(name: &str, fields: &[(&str, String)]) {
+    let Some(file) = HEALTH_FILE.get_or_init(open_health_file) else {
+        return;
+    };
+    write_event(file, name, fields);
+}
+
+pub fn default_health_log_path() -> PathBuf {
+    trace_root().join("health.log")
+}
+
+fn write_event(file: &Mutex<File>, name: &str, fields: &[(&str, String)]) {
     let now = SystemTime::now()
         .duration_since(UNIX_EPOCH)
         .map(|duration| duration.as_millis())
@@ -81,11 +98,49 @@ fn open_trace_file() -> Option<Mutex<File>> {
         .map(Mutex::new)
 }
 
+fn open_health_file() -> Option<Mutex<File>> {
+    let path = health_log_path_from_env()?;
+    rotate_health_log_if_needed(&path);
+    if let Some(parent) = path.parent() {
+        let _ = std::fs::create_dir_all(parent);
+    }
+    OpenOptions::new()
+        .create(true)
+        .append(true)
+        .open(&path)
+        .ok()
+        .map(Mutex::new)
+}
+
+fn health_log_path_from_env() -> Option<PathBuf> {
+    match std::env::var_os("DOSH_HEALTH_LOG") {
+        Some(raw) => {
+            let normalized = raw.to_string_lossy().to_ascii_lowercase();
+            if normalized.is_empty() || matches!(normalized.as_str(), "0" | "false" | "off") {
+                None
+            } else if matches!(normalized.as_str(), "1" | "true" | "on") {
+                Some(default_health_log_path())
+            } else {
+                Some(PathBuf::from(raw))
+            }
+        }
+        None => Some(default_health_log_path()),
+    }
+}
+
+fn rotate_health_log_if_needed(path: &Path) {
+    let Ok(metadata) = std::fs::metadata(path) else {
+        return;
+    };
+    if metadata.len() <= HEALTH_LOG_MAX_BYTES {
+        return;
+    }
+    let rotated = path.with_extension("log.1");
+    let _ = std::fs::remove_file(&rotated);
+    let _ = std::fs::rename(path, rotated);
+}
+
 fn default_trace_path() -> PathBuf {
-    let root = dirs::cache_dir()
-        .unwrap_or_else(std::env::temp_dir)
-        .join("dosh")
-        .join("trace");
     let exe = std::env::current_exe()
         .ok()
         .and_then(|path| {
@@ -93,7 +148,14 @@ fn default_trace_path() -> PathBuf {
                 .map(|stem| stem.to_string_lossy().to_string())
         })
         .unwrap_or_else(|| "dosh".to_string());
-    root.join(format!("{}-{}.log", exe, std::process::id()))
+    trace_root().join(format!("{}-{}.log", exe, std::process::id()))
+}
+
+fn trace_root() -> PathBuf {
+    dirs::cache_dir()
+        .unwrap_or_else(std::env::temp_dir)
+        .join("dosh")
+        .join("trace")
 }
 
 fn truthy_env(name: &str) -> bool {