fix(capture): retry capture v1 requests that get 404 or 429

With capture_mode="v1", 404 and 429 now retry like 408 and the 5xx
statuses, with the same backoff, Retry-After handling and retry budget.
They raise CaptureV1Error with their status after the last attempt.
This matches the v1 branch (#1132) and the harness capture v1
retry_behavior suite.

Author: andehen

.sampo/changesets/capture-v1-retry-404-429.md

@@ -0,0 +1,5 @@
+---
+pypi/posthog: patch
+---
+
+With `capture_mode="v1"`, capture now retries `404` and `429` responses like `408` and `500`, `502`, `503`, `504`, with the same backoff and `Retry-After` handling. Before, both failed the batch at once.

AGENTS.md

@@ -24,7 +24,7 @@ Follow [Public API changes](./CONTRIBUTING.md#public-api-changes). As an agent,
 
 Before changing capture configuration, serialization, routing, or retries, read the relevant implementation and tests.
 
-Preserve v0 defaults/compatibility; strictly typed v1 options and `$set`/`$set_once` relocation; v1-only compression (zlib-wrapped deflate, optional zstd); partial-only per-event retries with stable identity; accumulated drop reporting even on 2xx; terminal v1 `429`; `Retry-After` as a minimum bounded by the shared 30s ceiling; and inline blocking retries with `sync_mode=True`.
+Preserve v0 defaults/compatibility; strictly typed v1 options and `$set`/`$set_once` relocation; v1-only compression (zlib-wrapped deflate, optional zstd); partial-only per-event retries with stable identity; accumulated drop reporting even on 2xx; v1 retries on `404` and `429` like `408` and 5xx; `Retry-After` as a minimum bounded by the shared 30s ceiling; and inline blocking retries with `sync_mode=True`.
 
 ## Mirror and build safety
 

posthog/capture_v1.py

@@ -81,10 +81,9 @@
 _RESULT_DROP = "drop"
 _RESULT_RETRY = "retry"
 
-# HTTP status classification. 429 is terminal in v1 (unlike v0, where it is
-# retried) — the backend signals overload via retryable 5xx + Retry-After.
-_RETRYABLE_STATUSES = frozenset({408, 500, 502, 503, 504})
-_TERMINAL_STATUSES = frozenset({400, 401, 402, 413, 415, 429})
+# HTTP status classification. 404 and 429 can come from a proxy in front of capture.
+_RETRYABLE_STATUSES = frozenset({404, 408, 429, 500, 502, 503, 504})
+_TERMINAL_STATUSES = frozenset({400, 401, 402, 413, 415})
 
 # Single ceiling (seconds) for the retry backoff: caps the exponential schedule
 # and clamps a server ``Retry-After`` to the same value. Keeps the max retry
@@ -591,7 +590,7 @@ def _send_v1_batch(
             continue
 
         # Non-2xx. Retryable transient statuses back off; everything else
-        # (400/401/402/413/415/429/...) is terminal. Any drops collected from a
+        # (400/401/402/413/415/...) is terminal. Any drops collected from a
         # prior 2xx attempt ride along so on_error still sees them.
         v1_error = CaptureV1Error(
             parsed.status_code,

posthog/test/test_capture_v1.py

@@ -664,7 +664,7 @@ def test_malformed_2xx_is_terminal(self) -> None:
         self.assertEqual(len(stub.calls), 1)
         self.assertEqual(exc.status, 200)
 
-    @parameterized.expand([("bad_request", 400), ("rate_limited", 429)])
+    @parameterized.expand([("bad_request", 400), ("billing_limit", 402)])
     def test_terminal_status_raises_immediately(self, _name, status) -> None:
         stub, exc = self._run_expecting_error(
             [_msg("u-1")],
@@ -674,23 +674,29 @@ def test_terminal_status_raises_immediately(self, _name, status) -> None:
         self.assertEqual(len(stub.calls), 1)  # not retried
         self.assertEqual(exc.status, status)
 
-    def test_retryable_status_then_success(self) -> None:
+    @parameterized.expand(
+        [("not_found", 404), ("rate_limited", 429), ("unavailable", 503)]
+    )
+    def test_retryable_status_then_success(self, _name, status) -> None:
         stub = self._run(
             [_msg("u-1")],
             [
-                _FakeResponse(503, headers={"Retry-After": "2"}),
+                _FakeResponse(status, headers={"Retry-After": "2"}),
                 _results_response({"u-1": "ok"}),
             ],
         )
         self.assertEqual(len(stub.calls), 2)
         self.sleep.assert_called_once_with(2.0)  # honored Retry-After
 
-    def test_retryable_status_exhausted_raises(self) -> None:
+    @parameterized.expand(
+        [("not_found", 404), ("rate_limited", 429), ("unavailable", 503)]
+    )
+    def test_retryable_status_exhausted_raises(self, _name, status) -> None:
         stub, exc = self._run_expecting_error(
-            [_msg("u-1")], [_FakeResponse(503), _FakeResponse(503)], max_retries=1
+            [_msg("u-1")], [_FakeResponse(status), _FakeResponse(status)], max_retries=1
         )
         self.assertEqual(len(stub.calls), 2)
-        self.assertEqual(exc.status, 503)
+        self.assertEqual(exc.status, status)
 
     def test_transport_error_then_success(self) -> None:
         stub = self._run(