From b847c1ea5e72317bfbc55448066946b1bcb1b56e Mon Sep 17 00:00:00 2001
From: teknium1 <127238744+teknium1@users.noreply.github.com>
Date: Mon, 14 Sep 2026 18:49:46 -0700
Subject: [PATCH] fix(logging): name an unavailable log file once instead of
 silently dropping records

Builds on the salvaged EIO suppression: the reporter asked that logging
"degrade gracefully and identify the affected path". Print one stderr line
naming the file and errno when the stream first fails, reset the flag when
`_open()` succeeds again so a later failure is reported anew. The contributor's
test is replaced by one invariant: five records through a stream raising EIO
produce zero tracebacks and exactly one path mention, and the next emit
recovers into the real file.
---
 hermes_logging.py            | 13 ++++++++++---
 tests/test_hermes_logging.py |  2 +-
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/hermes_logging.py b/hermes_logging.py
index b1f2ffa8a3..5461256c1f 100644
--- a/hermes_logging.py
+++ b/hermes_logging.py
@@ -267,6 +267,7 @@ class _ManagedRotatingFileHandler(RotatingFileHandler):
     def __init__(self, *args, **kwargs):
         from hermes_cli.config import is_managed
         self._managed = is_managed()
+        self._unavailable_reported = False
         super().__init__(*args, **kwargs)
         self._record_stream_stat()
 
@@ -336,9 +337,14 @@ class _ManagedRotatingFileHandler(RotatingFileHandler):
         if _is_windows_concurrent_log_lock_timeout(exc):
             return
         if _is_unavailable_log_stream(exc):
-            # The QueueListener must not turn a transient filesystem failure
-            # into a traceback for every queued record. Drop the stale stream;
-            # the next emit will reopen it if the destination has recovered.
+            # The QueueListener must not turn a failing log destination into a traceback for
+            # every queued record. Name the path once, drop the stale stream; the next emit
+            # reopens it if the destination has recovered.
+            if not self._unavailable_reported:
+                self._unavailable_reported = True
+                _quietly(lambda: print(
+                    f"hermes_logging: {self.baseFilename} unavailable ({exc}); "
+                    "file logging paused until it recovers", file=_safe_stderr()))
             if self.stream is not None:
                 _quietly(self.stream.close)
             self.stream = None  # type: ignore[assignment]
@@ -348,6 +354,7 @@ class _ManagedRotatingFileHandler(RotatingFileHandler):
     def _open(self):
         stream = super()._open()
         self._chmod_if_managed()
+        self._unavailable_reported = False  # recovered: report again if it breaks anew
         return stream
 
     def doRollover(self):
diff --git a/tests/test_hermes_logging.py b/tests/test_hermes_logging.py
index 6a2073ea8b..0b2f6cdc6f 100644
--- a/tests/test_hermes_logging.py
+++ b/tests/test_hermes_logging.py
@@ -613,7 +613,7 @@ def test_eio_from_file_handler_names_the_path_once_then_recovers(tmp_path, capsy
         def writable(self):
             return True
 
-        def write(self, _s):
+        def write(self, *_a):
             raise OSError(5, "Input/output error")
 
         seek = tell = flush = write