Use tenacity instead of implementing a retry function ourselves (#1272)

Fixes #1234

.isort.cfg

@@ -4,4 +4,4 @@ include_trailing_comma=True
 force_grid_wrap=0
 use_parentheses=True
 line_length=88
-known_third_party = apispec,apispec_webframeworks,cerberus,dateutil,flask,flask_cors,hglib,imblearn,joblib,jsone,jsonschema,libmozdata,lmdb,marshmallow,matplotlib,microannotate,models,numpy,orjson,pandas,pkg_resources,pydriller,pyemd,pytest,redis,requests,responses,rq,scipy,setuptools,shap,sklearn,tabulate,taskcluster,tqdm,xgboost,yaml,zstandard
+known_third_party = apispec,apispec_webframeworks,cerberus,dateutil,flask,flask_cors,hglib,imblearn,joblib,jsone,jsonschema,libmozdata,lmdb,marshmallow,matplotlib,microannotate,models,numpy,orjson,pandas,pkg_resources,pydriller,pyemd,pytest,redis,requests,responses,rq,scipy,setuptools,shap,sklearn,tabulate,taskcluster,tenacity,tqdm,xgboost,yaml,zstandard

bugbug/utils.py

@@ -9,7 +9,6 @@ import logging
 import os
 import socket
 import tarfile
-import time
 from collections import deque
 from contextlib import contextmanager
 
@@ -204,18 +203,6 @@ def download_and_load_model(model_name):
     return get_model_class(model_name).load(path)
 
 
-def retry(operation, retries=5, wait_between_retries=30):
-    while True:
-        try:
-            return operation()
-        except Exception:
-            retries -= 1
-            if retries == 0:
-                raise
-
-            time.sleep(wait_between_retries)
-
-
 def zstd_compress(path):
     cctx = zstandard.ZstdCompressor()
     with open(path, "rb") as input_f:

requirements.txt

@@ -18,6 +18,7 @@ scipy==1.4.1
 shap[plots]==0.34.0
 tabulate==0.8.6
 taskcluster==24.2.0
+tenacity==6.0.0
 tqdm==4.42.1
 xgboost==0.90
 zstandard==0.13.0

scripts/commit_classifier.py

@@ -19,6 +19,7 @@ import matplotlib
 import numpy as np
 import requests
 import shap
+import tenacity
 from dateutil.relativedelta import relativedelta
 from libmozdata import vcs_map
 from libmozdata.phabricator import PhabricatorAPI
@@ -29,7 +30,6 @@ from bugbug.utils import (
     download_and_load_model,
     download_check_etag,
     get_secret,
-    retry,
     to_array,
     zstd_decompress,
 )
@@ -189,26 +189,32 @@ class CommitClassifier(object):
         logger.info(f"Cloning {repo_url}...")
 
         if not os.path.exists(repo_dir):
-            retry(
+            tenacity.retry(
                 lambda: subprocess.run(
                     ["git", "clone", "--quiet", repo_url, repo_dir], check=True
-                )
-            )
+                ),
+                wait=tenacity.wait_fixed(30),
+                stop=tenacity.stop_after_attempt(5),
+            )()
 
-        retry(
+        tenacity.retry(
             lambda: subprocess.run(
                 ["git", "pull", "--quiet", repo_url, "master"],
                 cwd=repo_dir,
                 capture_output=True,
                 check=True,
-            )
-        )
+            ),
+            wait=tenacity.wait_fixed(30),
+            stop=tenacity.stop_after_attempt(5),
+        )()
 
-        retry(
+        tenacity.retry(
             lambda: subprocess.run(
                 ["git", "checkout", rev], cwd=repo_dir, capture_output=True, check=True
-            )
-        )
+            ),
+            wait=tenacity.wait_fixed(30),
+            stop=tenacity.stop_after_attempt(5),
+        )()
 
     def update_commit_db(self):
         repository.clone(self.repo_dir)

scripts/microannotate_generator.py

@@ -5,10 +5,11 @@ import os
 import subprocess
 from logging import INFO, basicConfig, getLogger
 
+import tenacity
 from microannotate import generator
 
 from bugbug import db, repository
-from bugbug.utils import ThreadPoolExecutorResult, get_secret, retry
+from bugbug.utils import ThreadPoolExecutorResult, get_secret
 
 basicConfig(level=INFO)
 logger = getLogger(__name__)
@@ -61,11 +62,13 @@ class MicroannotateGenerator(object):
             else:
                 executor.submit(self.init_git_repo)
 
-        retry(
+        tenacity.retry(
             lambda: subprocess.run(
                 ["git", "config", "--global", "http.postBuffer", "12M"], check=True
-            )
-        )
+            ),
+            wait=tenacity.wait_fixed(30),
+            stop=tenacity.stop_after_attempt(5),
+        )()
 
         push_args = ["git", "push", repo_push_url, "master"]
         if is_old_version:
@@ -81,7 +84,11 @@ class MicroannotateGenerator(object):
                 remove_comments=self.remove_comments,
             )
 
-            retry(lambda: subprocess.run(push_args, cwd=self.git_repo_path, check=True))
+            tenacity.retry(
+                lambda: subprocess.run(push_args, cwd=self.git_repo_path, check=True),
+                wait=tenacity.wait_fixed(30),
+                stop=tenacity.stop_after_attempt(5),
+            )()
 
     def init_git_repo(self):
         subprocess.run(["git", "init", self.git_repo_path], check=True)
@@ -93,22 +100,26 @@ class MicroannotateGenerator(object):
         )
 
     def clone_git_repo(self):
-        retry(
+        tenacity.retry(
             lambda: subprocess.run(
                 ["git", "clone", "--quiet", self.repo_url, self.git_repo_path],
                 check=True,
-            )
-        )
+            ),
+            wait=tenacity.wait_fixed(30),
+            stop=tenacity.stop_after_attempt(5),
+        )()
 
         try:
-            retry(
+            tenacity.retry(
                 lambda: subprocess.run(
                     ["git", "pull", "--quiet", self.repo_url, "master"],
                     cwd=self.git_repo_path,
                     capture_output=True,
                     check=True,
-                )
-            )
+                ),
+                wait=tenacity.wait_fixed(30),
+                stop=tenacity.stop_after_attempt(5),
+            )()
         except subprocess.CalledProcessError as e:
             # When the repo is empty.
             if b"Couldn't find remote ref master" in e.stdout:

scripts/regressor_finder.py

@@ -14,6 +14,7 @@ from logging import INFO, basicConfig, getLogger
 
 import dateutil.parser
 import hglib
+import tenacity
 from dateutil.relativedelta import relativedelta
 from libmozdata import vcs_map
 from microannotate import utils as microannotate_utils
@@ -26,7 +27,7 @@ from bugbug.models.regressor import (
     BUG_INTRODUCING_COMMITS_DB,
     TOKENIZED_BUG_INTRODUCING_COMMITS_DB,
 )
-from bugbug.utils import download_and_load_model, retry, zstd_compress
+from bugbug.utils import download_and_load_model, zstd_compress
 
 basicConfig(level=INFO)
 logger = getLogger(__name__)
@@ -83,19 +84,23 @@ class RegressorFinder(object):
 
     def clone_git_repo(self, repo_url, repo_dir):
         if not os.path.exists(repo_dir):
-            retry(
+            tenacity.retry(
                 lambda: subprocess.run(
                     ["git", "clone", "--quiet", repo_url, repo_dir], check=True
-                )
+                ),
+                wait=tenacity.wait_fixed(30),
+                stop=tenacity.stop_after_attempt(5),
             )
 
-        retry(
+        tenacity.retry(
             lambda: subprocess.run(
                 ["git", "pull", "--quiet", repo_url, "master"],
                 cwd=repo_dir,
                 capture_output=True,
                 check=True,
-            )
+            ),
+            wait=tenacity.wait_fixed(30),
+            stop=tenacity.stop_after_attempt(5),
         )
 
     def init_mapping(self):