kpd: add optional mirror support

README.md

@@ -39,6 +39,21 @@ poetry install
 poetry run python -m unittest
 ```
 
+### Mirror setup
+
+To make more efficient use of network bandwidth consider having a mirror of your target git tree
+under /mirror/ or something like that and set the configuration attribute "mirror_dir" variable to the
+path where to find possible git trees.
+
+If your git tree is a linux clone set the "linux_clone" to true. In that case, in case your
+target exact basename repo is not in in the mirror path, for example {{ mirror_dir }}/linux-subsystem.git
+then the extra fallback path of {{ mirror_dir }}/linux.git will be used as a reference target.
+
+A reference target mirror path is only used if it exists. The mirror takes effect by leveraging
+the git clone --reference option when cloning. Using this can save considerable bandwidth and
+space, allowing kpd to run on thing guests on a corporate environment with for example an NFS
+mount for local git trees on a network.
+
 ## Running
 ```
 poetry run python -m kernel_patches_daemon --config <config_path> --label-color configs/labels.json

configs/kpd.json

@@ -41,5 +41,7 @@
       "github_oauth_token": "<TOKEN>"
     }
   },
-  "base_directory": "/tmp/repos"
+  "base_directory": "/tmp/repos",
+  "mirror_dir": "/mirror/",
+  "linux_clone": true
 }

kernel_patches_daemon/branch_worker.py

@@ -547,6 +547,8 @@ def __init__(
         app_auth: Optional[Auth.AppInstallationAuth] = None,
         email: Optional[EmailConfig] = None,
         http_retries: Optional[int] = None,
+        linux_clone: bool = False,
+        mirror_dir: Optional[str] = None,
     ) -> None:
         super().__init__(
             repo_url=repo_url,
@@ -559,6 +561,8 @@ def __init__(
         self.email = email
 
         self.log_extractor = log_extractor
+        self.mirror_dir = mirror_dir
+        self.linux_clone = linux_clone
         self.ci_repo_url = ci_repo_url
         self.ci_repo_dir = _uniq_tmp_folder(ci_repo_url, ci_branch, base_directory)
         self.ci_branch = ci_branch
@@ -682,9 +686,24 @@ def do_sync(self) -> None:
     def full_sync(self, path: str, url: str, branch: str) -> git.Repo:
         logging.info(f"Doing full clone from {redact_url(url)}, branch: {branch}")
 
+        multi_opts: Optional[List[str]] = None
+        if self.mirror_dir:
+            upstream_name = os.path.basename(self.upstream_url)
+            reference_path = os.path.join(self.mirror_dir, upstream_name)
+            fallback = None
+            if self.linux_clone:
+                fallback = os.path.join(self.mirror_dir, "linux.git")
+            if not os.path.exists(reference_path) and fallback and os.path.exists(fallback):
+                reference_path = fallback
+            if os.path.exists(reference_path):
+                multi_opts = ["--reference", reference_path]
+
         with HistogramMetricTimer(git_clone_duration, {"branch": branch}):
             shutil.rmtree(path, ignore_errors=True)
-            repo = git.Repo.clone_from(url, path)
+            if multi_opts:
+                repo = git.Repo.clone_from(url, path, multi_options=multi_opts)
+            else:
+                repo = git.Repo.clone_from(url, path)
             _reset_repo(repo, f"origin/{branch}")
 
         git_clone_counter.add(1, {"branch": branch})

kernel_patches_daemon/config.py

@@ -171,6 +171,8 @@ class KPDConfig:
     branches: Dict[str, BranchConfig]
     tag_to_branch_mapping: Dict[str, List[str]]
     base_directory: str
+    mirror_dir: Optional[str] = None
+    linux_clone: bool = False
 
     @classmethod
     def from_json(cls, json: Dict) -> "KPDConfig":
@@ -203,6 +205,8 @@ def from_json(cls, json: Dict) -> "KPDConfig":
                 for name, json_config in json["branches"].items()
             },
             base_directory=json["base_directory"],
+            mirror_dir=json.get("mirror_dir"),
+            linux_clone=json.get("linux_clone", False),
         )
 
     @classmethod

kernel_patches_daemon/github_sync.py

@@ -114,6 +114,8 @@ def __init__(
                 ci_branch=branch_config.ci_branch,
                 log_extractor=_log_extractor_from_project(kpd_config.patchwork.project),
                 base_directory=kpd_config.base_directory,
+                mirror_dir=kpd_config.mirror_dir,
+                linux_clone=kpd_config.linux_clone,
                 http_retries=http_retries,
                 github_oauth_token=branch_config.github_oauth_token,
                 app_auth=github_app_auth_from_branch_config(branch_config),

tests/test_branch_worker.py

@@ -68,6 +68,7 @@
 TEST_CI_REPO_URL = f"https://user:[email protected]/ci-org/{TEST_CI_REPO}"
 TEST_CI_BRANCH = "test_ci_branch"
 TEST_BASE_DIRECTORY = "/repos"
+TEST_MIRROR_DIRECTORY = "/mirror"
 TEST_BRANCH = "test-branch"
 TEST_CONFIG: Dict[str, Any] = {
     "version": 2,
@@ -124,6 +125,8 @@ def __init__(self, *args: Any, **kwargs: Any) -> None:
             "ci_branch": TEST_CI_BRANCH,
             "log_extractor": DefaultGithubLogExtractor(),
             "base_directory": TEST_BASE_DIRECTORY,
+            "mirror_dir": None,
+            "linux_clone": False,
         }
         presets.update(kwargs)
 
@@ -464,6 +467,56 @@ def test_fetch_repo_path_exists_git_exception(self) -> None:
             self._bw.fetch_repo(*fetch_params)
             fr.assert_called_once_with(*fetch_params)
 
+    def test_full_sync_with_mirror_dir(self) -> None:
+        bw = BranchWorkerMock(mirror_dir=TEST_MIRROR_DIRECTORY)
+        reference = os.path.join(
+            TEST_MIRROR_DIRECTORY, os.path.basename(TEST_UPSTREAM_REPO_URL)
+        )
+        with (
+            patch("kernel_patches_daemon.branch_worker.os.path.exists") as exists,
+            patch("kernel_patches_daemon.branch_worker.shutil.rmtree") as rm,
+        ):
+            exists.side_effect = lambda p: p == reference
+            bw.upstream_url = TEST_UPSTREAM_REPO_URL
+            bw.full_sync("somepath", "giturl", "branch")
+            self._git_repo_mock.clone_from.assert_called_once_with(
+                "giturl",
+                "somepath",
+                multi_options=["--reference", reference],
+            )
+
+    def test_full_sync_with_linux_mirror_fallback(self) -> None:
+        bw = BranchWorkerMock(mirror_dir=TEST_MIRROR_DIRECTORY, linux_clone=True)
+        fallback = os.path.join(TEST_MIRROR_DIRECTORY, "linux.git")
+        with (
+            patch("kernel_patches_daemon.branch_worker.os.path.exists") as exists,
+            patch("kernel_patches_daemon.branch_worker.shutil.rmtree") as rm,
+        ):
+            exists.side_effect = lambda p: p == fallback
+            bw.upstream_url = TEST_UPSTREAM_REPO_URL
+            bw.full_sync("somepath", "giturl", "branch")
+            self._git_repo_mock.clone_from.assert_called_once_with(
+                "giturl",
+                "somepath",
+                multi_options=["--reference", fallback],
+            )
+
+    def test_full_sync_without_linux_mirror_fallback(self) -> None:
+        bw = BranchWorkerMock(mirror_dir=TEST_MIRROR_DIRECTORY, linux_clone=False)
+        fallback = os.path.join(TEST_MIRROR_DIRECTORY, "linux.git")
+        with (
+            patch("kernel_patches_daemon.branch_worker.os.path.exists") as exists,
+            patch("kernel_patches_daemon.branch_worker.shutil.rmtree") as rm,
+        ):
+            exists.side_effect = lambda p: p == fallback
+            bw.upstream_url = TEST_UPSTREAM_REPO_URL
+            bw.full_sync("somepath", "giturl", "branch")
+            # Without linux_mirror we should not use fallback
+            self._git_repo_mock.clone_from.assert_called_once_with(
+                "giturl",
+                "somepath",
+            )
+
     def test_expire_branches(self) -> None:
         """Only the branch that matches pattern and is expired should be deleted"""
         not_expired_time = datetime.fromtimestamp(3 * BRANCH_TTL)

tests/test_config.py

@@ -208,5 +208,18 @@ def test_valid(self) -> None:
                 ),
             },
             base_directory="/repos",
+            mirror_dir=None,
+            linux_clone=False,
         )
         self.assertEqual(config, expected_config)
+
+    def test_linux_clone_enabled(self) -> None:
+        kpd_config_json = read_fixture("fixtures/kpd_config.json")
+        kpd_config_json["linux_clone"] = True
+
+        with patch(
+            "builtins.open", mock_open(read_data="TEST_KEY_FILE_CONTENT")
+        ):
+            config = KPDConfig.from_json(kpd_config_json)
+
+        self.assertTrue(config.linux_clone)

tests/test_github_sync.py

@@ -126,6 +126,20 @@ class TestCase:
                     gh.workers[TEST_BRANCH].ci_repo_dir.startswith(case.prefix),
                 )
 
+    def test_init_with_mirror_dir(self) -> None:
+        config = copy.copy(TEST_CONFIG)
+        config["mirror_dir"] = "/mirror"
+        kpd_config = KPDConfig.from_json(config)
+        gh = GithubSyncMock(kpd_config=kpd_config)
+        self.assertEqual("/mirror", gh.workers[TEST_BRANCH].mirror_dir)
+
+    def test_init_with_linux_clone(self) -> None:
+        config = copy.copy(TEST_CONFIG)
+        config["linux_clone"] = True
+        kpd_config = KPDConfig.from_json(config)
+        gh = GithubSyncMock(kpd_config=kpd_config)
+        self.assertTrue(gh.workers[TEST_BRANCH].linux_clone)
+
     def test_close_existing_prs_for_series(self) -> None:
         matching_pr_mock = MagicMock()
         matching_pr_mock.title = "matching"