diff --git a/‎docs/api-usage.rst‎
Lines changed: 18 additions & 0 deletions b/‎docs/api-usage.rst‎
Lines changed: 18 additions & 0 deletions
diff --git a/‎gitlab/__init__.py‎
Lines changed: 5 additions & 1 deletion b/‎gitlab/__init__.py‎
Lines changed: 5 additions & 1 deletion
@@ -348,3 +348,21 @@ throttled, you can set this parameter to -1. This parameter is ignored if
 .. warning::
 
    You will get an Exception, if you then go over the rate limit of your GitLab instance.
+
+Transient errors
+----------------
+
+GitLab server can sometimes return a transient HTTP error.
+python-gitlab can automatically retry in such case, when
+``retry_transient_errors`` argument is set to ``True``.  When enabled,
+HTTP error codes 500 (Internal Server Error), 502 (502 Bad Gateway),
+503 (Service Unavailable), and 504 (Gateway Timeout) are retried.  By
+default an exception is raised for these errors.
+
+.. code-block:: python
+
+   import gitlab
+   import requests
+
+   gl = gitlab.gitlab(url, token, api_version=4)
+   gl.projects.list(all=True, retry_transient_errors=True)
@@ -518,6 +518,8 @@ def http_request(
 
         # obey the rate limit by default
         obey_rate_limit = kwargs.get("obey_rate_limit", True)
+        # do not retry transient errors by default
+        retry_transient_errors = kwargs.get("retry_transient_errors", False)
 
         # set max_retries to 10 by default, disable by setting it to -1
         max_retries = kwargs.get("max_retries", 10)
@@ -531,7 +533,9 @@ def http_request(
             if 200 <= result.status_code < 300:
                 return result
 
-            if 429 == result.status_code and obey_rate_limit:
+            if (429 == result.status_code and obey_rate_limit) or (
+                result.status_code in [500, 502, 503, 504] and retry_transient_errors
+            ):
                 if max_retries == -1 or cur_retries < max_retries:
                     wait_time = 2 ** cur_retries * 0.1
                     if "Retry-After" in result.headers: