redis-benchmarks-specification 0.1.337__tar.gz → 0.1.339__tar.gz

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: redis-benchmarks-specification
-Version: 0.1.337
+Version: 0.1.339
 Summary: The Redis benchmarks specification describes the cross-language/tools requirements and expectations to foster performance and observability standards around redis related technologies. Members from both industry and academia, including organizations and individuals are encouraged to contribute.
 Author: filipecosta90
 Author-email: filipecosta.90@gmail.com

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "redis-benchmarks-specification"
-version = "0.1.337"
+version = "0.1.339"
 description = "The Redis benchmarks specification describes the cross-language/tools requirements and expectations to foster performance and observability standards around redis related technologies. Members from both industry and academia, including organizations and individuals are encouraged to contribute."
 authors = ["filipecosta90 <filipecosta.90@gmail.com>","Redis Performance Group <performance@redis.com>"]
 readme = "Readme.md"

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/redis_benchmarks_specification/__builder__/builder.py

@@ -429,6 +429,79 @@ def builder_process_stream(
                 if b"server_name" in testDetails:
                     server_name = testDetails[b"server_name"].decode()
 
+                # Check if artifacts already exist before building
+                prefix = f"github_org={github_org}/github_repo={github_repo}/git_branch={str(git_branch)}/git_version={str(git_version)}/git_hash={str(git_hash)}"
+
+                # Create a comprehensive build signature that includes all build-affecting parameters
+                import hashlib
+
+                build_signature_parts = [
+                    str(id),  # build config ID
+                    str(build_command),  # build command
+                    str(build_vars_str),  # environment variables
+                    str(compiler),  # compiler
+                    str(cpp_compiler),  # C++ compiler
+                    str(build_image),  # build image
+                    str(build_os),  # OS
+                    str(build_arch),  # architecture
+                    ",".join(sorted(build_artifacts)),  # artifacts list
+                ]
+                build_signature = hashlib.sha256(
+                    ":".join(build_signature_parts).encode()
+                ).hexdigest()[:16]
+
+                # Check if all artifacts already exist
+                all_artifacts_exist = True
+                artifact_keys = {}
+                for artifact in build_artifacts:
+                    bin_key = f"zipped:artifacts:{prefix}:{id}:{build_signature}:{artifact}.zip"
+                    artifact_keys[artifact] = bin_key
+                    if not conn.exists(bin_key):
+                        all_artifacts_exist = False
+                        break
+
+                if all_artifacts_exist:
+                    logging.info(
+                        f"Artifacts for {git_hash}:{id} with build signature {build_signature} already exist, reusing them"
+                    )
+                    # Skip build and reuse existing artifacts
+                    build_stream_fields, result = generate_benchmark_stream_request(
+                        id,
+                        conn,
+                        run_image,
+                        build_arch,
+                        testDetails,
+                        build_os,
+                        build_artifacts,
+                        build_command,
+                        build_config_metadata,
+                        build_image,
+                        build_vars_str,
+                        compiler,
+                        cpp_compiler,
+                        git_branch,
+                        git_hash,
+                        git_timestamp_ms,
+                        git_version,
+                        pull_request,
+                        None,  # redis_temporary_dir not needed for reuse
+                        tests_groups_regexp,
+                        tests_priority_lower_limit,
+                        tests_priority_upper_limit,
+                        tests_regexp,
+                        ".*",  # command_regexp - default to all commands
+                        use_git_timestamp,
+                        server_name,
+                        github_org,
+                        github_repo,
+                        artifact_keys,  # Pass existing artifact keys
+                    )
+                    continue  # Skip to next build spec
+
+                logging.info(
+                    f"Building artifacts for {git_hash}:{id} with build signature {build_signature}"
+                )
+
                 build_start_datetime = datetime.datetime.utcnow()
                 logging.info(
                     "Using the following build command {}.".format(build_command)
@@ -507,6 +580,7 @@ def builder_process_stream(
                     server_name,
                     github_org,
                     github_repo,
+                    None,  # existing_artifact_keys - None for new builds
                 )
                 if result is True:
                     benchmark_stream_id = conn.xadd(
@@ -648,6 +722,7 @@ def generate_benchmark_stream_request(
     server_name="redis",
     github_org="redis",
     github_repo="redis",
+    existing_artifact_keys=None,
 ):
     build_stream_fields = {
         "id": id,
@@ -691,21 +766,50 @@ def generate_benchmark_stream_request(
     if git_timestamp_ms is not None:
         build_stream_fields["git_timestamp_ms"] = git_timestamp_ms
 
-    prefix = f"github_org={github_org}/github_repo={github_repo}/git_branch={str(git_branch)}/git_version={str(git_version)}/git_hash={str(git_hash)}"
-    for artifact in build_artifacts:
-        bin_key = f"zipped:artifacts:{prefix}:{id}:{artifact}.zip"
-        if artifact == "redisearch.so":
-            bin_artifact = open(
-                f"{redis_temporary_dir}modules/redisearch/src/bin/linux-x64-release/search-community/{artifact}",
-                "rb",
-            ).read()
-        else:
-            bin_artifact = open(f"{redis_temporary_dir}src/{artifact}", "rb").read()
-        bin_artifact_len = len(bytes(bin_artifact))
-        assert bin_artifact_len > 0
-        conn.set(bin_key, bytes(bin_artifact), ex=REDIS_BINS_EXPIRE_SECS)
-        build_stream_fields[artifact] = bin_key
-        build_stream_fields["{}_len_bytes".format(artifact)] = bin_artifact_len
+    if existing_artifact_keys is not None:
+        # Use existing artifact keys (for reuse case)
+        for artifact in build_artifacts:
+            bin_key = existing_artifact_keys[artifact]
+            build_stream_fields[artifact] = bin_key
+            # Get the length from the existing artifact
+            bin_artifact_len = conn.strlen(bin_key)
+            build_stream_fields["{}_len_bytes".format(artifact)] = bin_artifact_len
+    else:
+        # Build new artifacts and store them
+        prefix = f"github_org={github_org}/github_repo={github_repo}/git_branch={str(git_branch)}/git_version={str(git_version)}/git_hash={str(git_hash)}"
+
+        # Create build signature for new artifacts
+        import hashlib
+
+        build_signature_parts = [
+            str(id),  # build config ID
+            str(build_command),  # build command
+            str(build_vars_str),  # environment variables
+            str(compiler),  # compiler
+            str(cpp_compiler),  # C++ compiler
+            str(build_image),  # build image
+            str(build_os),  # OS
+            str(build_arch),  # architecture
+            ",".join(sorted(build_artifacts)),  # artifacts list
+        ]
+        build_signature = hashlib.sha256(
+            ":".join(build_signature_parts).encode()
+        ).hexdigest()[:16]
+
+        for artifact in build_artifacts:
+            bin_key = f"zipped:artifacts:{prefix}:{id}:{build_signature}:{artifact}.zip"
+            if artifact == "redisearch.so":
+                bin_artifact = open(
+                    f"{redis_temporary_dir}modules/redisearch/src/bin/linux-x64-release/search-community/{artifact}",
+                    "rb",
+                ).read()
+            else:
+                bin_artifact = open(f"{redis_temporary_dir}src/{artifact}", "rb").read()
+            bin_artifact_len = len(bytes(bin_artifact))
+            assert bin_artifact_len > 0
+            conn.set(bin_key, bytes(bin_artifact), ex=REDIS_BINS_EXPIRE_SECS)
+            build_stream_fields[artifact] = bin_key
+            build_stream_fields["{}_len_bytes".format(artifact)] = bin_artifact_len
     result = True
     if b"platform" in testDetails:
         build_stream_fields["platform"] = testDetails[b"platform"]

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/redis_benchmarks_specification/__cli__/cli.py

@@ -104,6 +104,12 @@ def trigger_tests_dockerhub_cli_command_logic(args, project_name, project_versio
         0,
         10000,
         args.tests_regexp,
+        ".*",  # command_regexp
+        False,  # use_git_timestamp
+        "redis",  # server_name
+        "redis",  # github_org
+        "redis",  # github_repo
+        None,  # existing_artifact_keys
     )
     build_stream_fields["github_repo"] = args.gh_repo
     build_stream_fields["github_org"] = args.gh_org

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/redis_benchmarks_specification/__runner__/runner.py

@@ -1934,7 +1934,7 @@ def process_self_contained_coordinator_stream(
                                     benchmark_tool_global=benchmark_tool_global,
                                 )
                                 continue
-                            logging.info(
+                            logging.debug(
                                 "Test {} priority ({}) is within the priority limit [{},{}]".format(
                                     test_name,
                                     priority,

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/redis_benchmarks_specification/__self_contained_coordinator__/args.py

@@ -205,6 +205,6 @@ def create_self_contained_coordinator_args(project_name):
         "--skip-clear-pending-on-startup",
         default=False,
         action="store_true",
-        help="Skip automatically clearing pending messages for this consumer on startup. By default, pending messages are cleared to recover from crashes.",
+        help="Skip automatically clearing pending messages and resetting consumer group position on startup. By default, pending messages are cleared and consumer group is reset to latest position to skip old work and recover from crashes.",
     )
     return parser

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/redis_benchmarks_specification/__self_contained_coordinator__/runners.py

@@ -143,6 +143,26 @@ def clear_pending_messages_for_consumer(conn, running_platform, consumer_pos):
         logging.error(f"Unexpected error clearing pending messages: {e}")
 
 
+def reset_consumer_group_to_latest(conn, running_platform):
+    """Reset the consumer group position to only read new messages (skip old ones)"""
+    consumer_group_name = get_runners_consumer_group_name(running_platform)
+
+    try:
+        # Set the consumer group position to '$' (latest) to skip all existing messages
+        conn.xgroup_setid(STREAM_KEYNAME_NEW_BUILD_EVENTS, consumer_group_name, id="$")
+        logging.info(
+            f"Reset consumer group {consumer_group_name} position to latest - will only process new messages"
+        )
+
+    except redis.exceptions.ResponseError as e:
+        if "NOGROUP" in str(e):
+            logging.info(f"Consumer group {consumer_group_name} does not exist yet")
+        else:
+            logging.warning(f"Error resetting consumer group position: {e}")
+    except Exception as e:
+        logging.error(f"Unexpected error resetting consumer group position: {e}")
+
+
 def process_self_contained_coordinator_stream(
     conn,
     datasink_push_results_redistimeseries,

{redis_benchmarks_specification-0.1.337 → redis_benchmarks_specification-0.1.339}/redis_benchmarks_specification/__self_contained_coordinator__/self_contained_coordinator.py

@@ -76,6 +76,7 @@ from redis_benchmarks_specification.__self_contained_coordinator__.runners impor
     build_runners_consumer_group_create,
     get_runners_consumer_group_name,
     clear_pending_messages_for_consumer,
+    reset_consumer_group_to_latest,
 )
 from redis_benchmarks_specification.__setups__.topologies import get_topologies
 
@@ -651,15 +652,20 @@ def main():
     running_platform = args.platform_name
     build_runners_consumer_group_create(gh_event_conn, running_platform)
 
-    # Clear pending messages by default (unless explicitly skipped)
+    # Clear pending messages and reset consumer group position by default (unless explicitly skipped)
     if not args.skip_clear_pending_on_startup:
         consumer_pos = args.consumer_pos
-        logging.info("Clearing pending messages on startup (default behavior)")
+        logging.info(
+            "Clearing pending messages and resetting consumer group position on startup (default behavior)"
+        )
         clear_pending_messages_for_consumer(
             gh_event_conn, running_platform, consumer_pos
         )
+        reset_consumer_group_to_latest(gh_event_conn, running_platform)
     else:
-        logging.info("Skipping pending message cleanup as requested")
+        logging.info(
+            "Skipping pending message cleanup and consumer group reset as requested"
+        )
 
     stream_id = None
     docker_client = docker.from_env()