Cleanup script - update timeout (#3387)

* draft of script

* remove private repos

* add cutoff to created repos

* update check for maintainer

* add warning for repos with critical amount of maintainers

* include comments on issues as being active

* add debug options, change to rest api to catch edge case

* performance improvements

* start script

* create PR from results

* improvements on grouping

* add repos to ignore

* use display name on readme PR

* fix activity per team and not per user

* add more ignore repos to the list

* fix role check per repo

* fix

* update usage text

* only run in a few repos for starters

* add timeout and retry

* dont retry on expected 404

Author: maryliag

scripts/move-to-emeritus.py

@@ -104,7 +104,9 @@ def get_cutoff_date():
     cutoff = now - timedelta(days=INACTIVITY_MONTHS * 30)
     return cutoff.strftime("%Y-%m-%d")
 
-def request_with_retry(method, url, data=None, retries=3):
+MAX_RATE_LIMIT_WAIT = 300  # cap waits at 5 min to avoid token expiry
+
+def request_with_retry(method, url, data=None, retries=5):
     for attempt in range(1, retries + 1):
         req = urllib.request.Request(url, headers=HEADERS, method=method)
         if data is not None:
@@ -113,7 +115,7 @@ def request_with_retry(method, url, data=None, retries=3):
         else:
             body = None
         try:
-            resp = urllib.request.urlopen(req, body)
+            resp = urllib.request.urlopen(req, body, timeout=30)
             return resp
         except urllib.error.HTTPError as e:
             if e.code in (403, 429):
@@ -125,12 +127,24 @@ def request_with_retry(method, url, data=None, retries=3):
                     wait = max(0, int(reset_time) - int(time.time())) + 1
                 else:
                     wait = 60
+                wait = min(wait, MAX_RATE_LIMIT_WAIT)
                 print(f"Rate limited. Waiting {wait}s before retry...")
                 time.sleep(wait)
                 continue
-            print(f"Attempt {attempt} failed: HTTP {e.code}")
+            if 500 <= e.code < 600:
+                # Transient server error — retry with backoff
+                print(f"Attempt {attempt} failed: HTTP {e.code}")
+                if attempt < retries:
+                    time.sleep(2 ** attempt)
+                else:
+                    raise
+                continue
+            # 4xx (other than 429/403): deterministic — don't retry
+            raise
+        except (TimeoutError, OSError) as e:
+            print(f"Attempt {attempt} failed: {e}")
             if attempt < retries:
-                time.sleep(1)
+                time.sleep(2)
             else:
                 raise
     return None