Merge pull request ClickHouse#63543 from ClickHouse/find-proper-commit-for-a-sync

Find a proper commit for cumulative `A Sync` status

Author: Felixoid

tests/ci/ci.py

@@ -32,11 +32,10 @@
     RerunHelper,
     format_description,
     get_commit,
-    get_commit_filtered_statuses,
     post_commit_status,
     set_status_comment,
-    trigger_mergeable_check,
     update_mergeable_check,
+    update_upstream_sync_status,
 )
 from digest_helper import DockerDigester, JobDigester
 from env_helper import (
@@ -55,7 +54,7 @@
 from git_helper import Runner as GitRunner
 from github_helper import GitHub
 from pr_info import PRInfo
-from report import ERROR, SUCCESS, BuildResult, JobReport, get_status
+from report import ERROR, SUCCESS, BuildResult, JobReport
 from s3_helper import S3Helper
 from synchronizer_utils import SYNC_BRANCH_PREFIX
 from version_helper import get_version_from_repo
@@ -2204,30 +2203,19 @@ def main() -> int:
                         and mergeable_status
                         and GITHUB_REPOSITORY != GITHUB_UPSTREAM_REPOSITORY
                     ):
-                        pr_number = int(pr_info.head_ref.split("/pr/", maxsplit=1)[1])
-                        upstream_repo = gh.get_repo(GITHUB_UPSTREAM_REPOSITORY)
-                        head_sha = upstream_repo.get_pull(pr_number).head.sha
-                        upstream_commit = upstream_repo.get_commit(head_sha)
-                        post_commit_status(
-                            upstream_commit,
-                            get_status(mergeable_status.state),
-                            "",  # let's won't expose any urls from cloud
-                            mergeable_status.description,
-                            StatusNames.SYNC,
+                        upstream_pr_number = int(
+                            pr_info.head_ref.split("/pr/", maxsplit=1)[1]
                         )
-                        trigger_mergeable_check(
-                            upstream_commit,
-                            get_commit_filtered_statuses(upstream_commit),
-                            True,
+                        update_upstream_sync_status(
+                            upstream_pr_number, pr_info.number, gh, mergeable_status
                         )
-
                         prepared_events = prepare_tests_results_for_clickhouse(
                             pr_info,
                             [],
                             job_report.status,
                             0,
                             job_report.start_time,
-                            f"https://github.com/ClickHouse/ClickHouse/pull/{pr_number}",
+                            f"https://github.com/ClickHouse/ClickHouse/pull/{upstream_pr_number}",
                             StatusNames.SYNC,
                         )
                         prepared_events[0]["test_context_raw"] = args.job_name

tests/ci/commit_status_helper.py

@@ -18,7 +18,12 @@
 from github.Repository import Repository
 
 from ci_config import CHECK_DESCRIPTIONS, REQUIRED_CHECKS, CheckDescription, StatusNames
-from env_helper import GITHUB_REPOSITORY, GITHUB_RUN_URL, TEMP_PATH
+from env_helper import (
+    GITHUB_REPOSITORY,
+    GITHUB_RUN_URL,
+    GITHUB_UPSTREAM_REPOSITORY,
+    TEMP_PATH,
+)
 from lambda_shared_package.lambda_shared.pr import Labels
 from pr_info import PRInfo
 from report import (
@@ -29,6 +34,7 @@
     StatusType,
     TestResult,
     TestResults,
+    get_status,
     get_worst_status,
 )
 from s3_helper import S3Helper
@@ -500,3 +506,60 @@ def trigger_mergeable_check(
         return set_mergeable_check(commit, description, state, hide_url)
 
     return mergeable_status
+
+
+def update_upstream_sync_status(
+    upstream_pr_number: int,
+    sync_pr_number: int,
+    gh: Github,
+    mergeable_status: CommitStatus,
+) -> None:
+    upstream_repo = gh.get_repo(GITHUB_UPSTREAM_REPOSITORY)
+    upstream_pr = upstream_repo.get_pull(upstream_pr_number)
+    sync_repo = gh.get_repo(GITHUB_REPOSITORY)
+    sync_pr = sync_repo.get_pull(sync_pr_number)
+    # Find the commit that is in both repos, upstream and cloud
+    sync_commits = sync_pr.get_commits().reversed
+    upstream_commits = upstream_pr.get_commits()
+    # Github objects are compared by _url attribute. We can't compare them directly and
+    # should compare commits by SHA1
+    upstream_shas = [uc.sha for uc in upstream_commits]
+    logging.info("Commits in upstream PR:\n %s", ", ".join(upstream_shas))
+    sync_shas = [uc.sha for uc in upstream_commits]
+    logging.info("Commits in sync PR:\n %s", ", ".join(reversed(sync_shas)))
+    found = False
+    for commit in sync_commits:
+        try:
+            idx = upstream_shas.index(commit.sha)
+            found = True
+            upstream_commit = upstream_commits[idx]
+            break
+        except ValueError:
+            continue
+
+    if not found:
+        logging.info(
+            "There's no same commits in upstream and sync PRs, probably force-push"
+        )
+        return
+
+    sync_status = get_status(mergeable_status.state)
+    logging.info(
+        "Using commit %s to post the %s status `%s`: [%s]",
+        upstream_commit.sha,
+        sync_status,
+        StatusNames.SYNC,
+        mergeable_status.description,
+    )
+    post_commit_status(
+        upstream_commit,
+        sync_status,
+        "",  # let's won't expose any urls from cloud
+        mergeable_status.description,
+        StatusNames.SYNC,
+    )
+    trigger_mergeable_check(
+        upstream_commit,
+        get_commit_filtered_statuses(upstream_commit),
+        True,
+    )