update: yarn_client.py add parsed html fallback

spark_executor/core/yarn_client.py

@@ -239,16 +239,21 @@ def _fetch_logs_via_am_container(application_id: str, config: YarnClientConfig)
 
     # Fallback path: Knox / wrapped YARN serves an HTML directory listing for
     # every URL on the NodeManager log endpoint. Parse it and fetch each file.
-    resp = _request("GET", am_container_logs, timeout=60.0, verify=verify, auth=auth)
-    if resp.status_code < 400:
-        parts: list[str] = []
-        for filename in _parse_log_listing_html(resp.text):
-            file_url = f"{am_container_logs}/{filename}"
-            file_resp = _request("GET", file_url, timeout=60.0, verify=verify, auth=auth)
-            if file_resp.status_code < 400 and not _looks_like_html(file_resp.text):
-                parts.append(f"=== {filename} ===\n{file_resp.text}")
-        if parts:
-            return "\n\n".join(parts)
+    orig_resp_text = resp.text
+    try:
+        resp = _request("GET", am_container_logs, timeout=60.0, verify=verify, auth=auth)
+        if resp.status_code < 400:
+            parts: list[str] = []
+            for filename in _parse_log_listing_html(resp.text):
+                file_url = f"{am_container_logs}/{filename}"
+                file_resp = _request("GET", file_url, timeout=60.0, verify=verify, auth=auth)
+                if file_resp.status_code < 400 and not _looks_like_html(file_resp.text):
+                    parts.append(f"=== {filename} ===\n{file_resp.text}")
+            if parts:
+                return "\n\n".join(parts)
+    except Exception as ex:
+        logger.warning(f"Parse html error: {ex}")
+        return orig_resp_text
 
     raise _logs_unavailable_error(application_id)