Builder cleanup, arch-specific streams, CLI/Runner/Compare/Coordinator enhancements, and new benchmark datasets (#312)

* Bumping version to 0.1.310

* Added gcc:8.5.0-arm64-debian-buster-default

* Moved from gcc-8 to gcc-15. runner image using debian:bookworm

* bumping version

* Removed reference to gcc-8 on tests

* Fixed missing gcc-8 removal

* renamed buster->bookworm

* Added extra session caching, RPUSH, and SET benchmarks

* working on box plot charts on compare tool

* Added command filter an extra benchmark for 1:10 set/get 1KB benchmark

* bumping version to 0.1.325

* added uri support for benchmark runner

* fixes for 0.1.334

* bumping version

* version 0.1.338

* per arch streams

* stash

* Removed dump.rdb

Author: fcostaoliveira

poetry.lock

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "redis-benchmarks-specification"
-version = "0.1.323"
+version = "0.2.11"
 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"
@@ -27,6 +27,7 @@ pandas = "^2.1.2"
 numpy = "^2.0.0"
 jsonpath-ng = "^1.6.1"
 
+seaborn = "^0.13.2"
 [tool.poetry.dev-dependencies]
 click = "8.1.7"
 black = "24.4.2"

pyproject.toml

@@ -138,7 +138,7 @@ def spec_cli_args(parser):
     parser.add_argument("--gh_repo", type=str, default="redis")
     parser.add_argument("--server_name", type=str, default=None)
     parser.add_argument("--run_image", type=str, default="redis")
-    parser.add_argument("--build_arch", type=str, default=None)
+    parser.add_argument("--arch", type=str, default="amd64")
     parser.add_argument("--id", type=str, default="dockerhub")
     parser.add_argument("--mnt_point", type=str, default="")
     parser.add_argument("--trigger-unstable-commits", type=bool, default=True)
@@ -217,4 +217,10 @@ def spec_cli_args(parser):
         default=-1,
         help="Wait x sections for build. If -1, waits forever.",
     )
+    parser.add_argument(
+        "--command-regex",
+        type=str,
+        default=".*",
+        help="Filter tests by command using regex. Only tests that include commands matching this regex will be processed.",
+    )
     return parser

redis_benchmarks_specification/__builder__/builder.py

@@ -44,6 +44,7 @@
     STREAM_KEYNAME_GH_EVENTS_COMMIT,
     STREAM_GH_EVENTS_COMMIT_BUILDERS_CG,
     STREAM_KEYNAME_NEW_BUILD_EVENTS,
+    get_arch_specific_stream_name,
 )
 from redis_benchmarks_specification.__common__.package import (
     get_version_string,
@@ -84,7 +85,7 @@ def trigger_tests_dockerhub_cli_command_logic(args, project_name, project_versio
         args.id,
         conn,
         args.run_image,
-        args.build_arch,
+        args.arch,
         testDetails,
         "n/a",
         [],
@@ -104,6 +105,12 @@ 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
@@ -118,9 +125,12 @@ def trigger_tests_dockerhub_cli_command_logic(args, project_name, project_versio
         store_airgap_image_redis(conn, docker_client, args.run_image)
 
     if result is True:
-        benchmark_stream_id = conn.xadd(
-            STREAM_KEYNAME_NEW_BUILD_EVENTS, build_stream_fields
+        # Use architecture-specific stream
+        arch_specific_stream = get_arch_specific_stream_name(args.arch)
+        logging.info(
+            f"CLI adding work to architecture-specific stream: {arch_specific_stream}"
         )
+        benchmark_stream_id = conn.xadd(arch_specific_stream, build_stream_fields)
         logging.info(
             "sucessfully requested a new run {}. Stream id: {}".format(
                 build_stream_fields, benchmark_stream_id
@@ -432,9 +442,9 @@ def trigger_tests_cli_command_logic(args, project_name, project_version):
             commit_dict["tests_groups_regexp"] = tests_groups_regexp
             commit_dict["github_org"] = args.gh_org
             commit_dict["github_repo"] = args.gh_repo
-            if args.build_arch is not None:
-                commit_dict["build_arch"] = args.build_arch
-                commit_dict["arch"] = args.build_arch
+            if args.arch is not None:
+                commit_dict["build_arch"] = args.arch
+                commit_dict["arch"] = args.arch
             if args.server_name is not None and args.server_name != "":
                 commit_dict["server_name"] = args.server_name
             if args.build_artifacts != "":

redis_benchmarks_specification/__cli__/args.py

@@ -32,6 +32,20 @@
     "STREAM_KEYNAME_NEW_BUILD_EVENTS", "oss:api:gh/redis/redis/builds"
 )
 
+
+# Function to get architecture-specific build events stream name
+def get_arch_specific_stream_name(arch):
+    """Get architecture-specific stream name for build events"""
+    base_stream = STREAM_KEYNAME_NEW_BUILD_EVENTS
+    if arch in ["amd64", "x86_64"]:
+        return f"{base_stream}:amd64"
+    elif arch in ["arm64", "aarch64"]:
+        return f"{base_stream}:arm64"
+    else:
+        # Fallback to base stream for unknown architectures
+        return base_stream
+
+
 STREAM_GH_NEW_BUILD_RUNNERS_CG = os.getenv(
     "STREAM_GH_NEW_BUILD_RUNNERS_CG", "runners-cg:redis/redis/commits"
 )

redis_benchmarks_specification/__cli__/cli.py

@@ -101,7 +101,7 @@ def get_benchmark_specs(testsuites_folder, test="", test_regex=".*"):
             for test_name in original_files:
                 match_obj = re.search(test_regexp_string, test_name)
                 if match_obj is None:
-                    logging.info(
+                    logging.debug(
                         "Skipping test file: {} given it does not match regex {}".format(
                             test_name, test_regexp_string
                         )
@@ -291,13 +291,7 @@ def export_redis_metrics(
             metric_name,
             metric_value,
         ) in overall_end_time_metrics.items():
-            tsname_metric = "{}/{}/{}/benchmark_end/{}/{}".format(
-                sprefix,
-                test_name,
-                by_variant,
-                setup_name,
-                metric_name,
-            )
+            tsname_metric = f"{sprefix}/{test_name}/{by_variant}/benchmark_end/{running_platform}/{setup_name}/{metric_name}"
 
             logging.debug(
                 "Adding a redis server side metric collected at the end of benchmark."
@@ -404,6 +398,7 @@ def exporter_datasink_common(
         running_platform,
         None,
         git_hash,
+        disable_target_tables=True,
     )
     if collect_memory_metrics:
         logging.info("Collecting memory metrics")

redis_benchmarks_specification/__common__/env.py

@@ -0,0 +1,20 @@
+"""
+Warning suppression module that should be imported first to suppress known warnings.
+"""
+
+import warnings
+
+# Suppress cryptography deprecation warnings from paramiko
+warnings.filterwarnings("ignore", category=DeprecationWarning, module="paramiko")
+warnings.filterwarnings("ignore", message=".*TripleDES.*", category=DeprecationWarning)
+warnings.filterwarnings(
+    "ignore", message=".*cryptography.*", category=DeprecationWarning
+)
+
+# Also suppress the specific CryptographyDeprecationWarning if it exists
+try:
+    from cryptography.utils import CryptographyDeprecationWarning
+
+    warnings.filterwarnings("ignore", category=CryptographyDeprecationWarning)
+except ImportError:
+    pass

redis_benchmarks_specification/__common__/runner.py

@@ -838,34 +838,47 @@ def common_exporter_logic(
         and artifact_version != ""
         and artifact_version != "N/A"
     ):
-        # extract per-version datapoints
-        total_hs_ts = len(per_hash_time_series_dict.keys())
-        logging.info(
-            f"Extending the by.hash {git_hash} timeseries ({total_hs_ts}) with version info {artifact_version}"
-        )
-        for hash_timeserie in per_hash_time_series_dict.values():
-            hash_timeserie["labels"]["version"] = artifact_version
-        (
-            _,
-            per_version_time_series_dict,
-            version_target_tables,
-        ) = extract_perversion_timeseries_from_results(
-            used_ts,
-            metrics,
-            results_dict,
-            artifact_version,
-            tf_github_org,
-            tf_github_repo,
-            deployment_name,
-            deployment_type,
-            test_name,
-            tf_triggering_env,
-            metadata_tags,
-            build_variant_name,
-            running_platform,
-            testcase_metric_context_paths,
-        )
-        total_break_by_added += 1
+        # Check if version 255.255.255 should only be pushed for unstable branch
+        should_push_version = True
+        if artifact_version == "255.255.255":
+            if tf_github_branch != "unstable":
+                logging.info(
+                    f"Skipping version 255.255.255 data push for branch '{tf_github_branch}' "
+                    f"(only pushing for 'unstable' branch)"
+                )
+                should_push_version = False
+            else:
+                logging.info(f"Pushing version 255.255.255 data for unstable branch")
+
+        if should_push_version:
+            # extract per-version datapoints
+            total_hs_ts = len(per_hash_time_series_dict.keys())
+            logging.info(
+                f"Extending the by.hash {git_hash} timeseries ({total_hs_ts}) with version info {artifact_version}"
+            )
+            for hash_timeserie in per_hash_time_series_dict.values():
+                hash_timeserie["labels"]["version"] = artifact_version
+            (
+                _,
+                per_version_time_series_dict,
+                version_target_tables,
+            ) = extract_perversion_timeseries_from_results(
+                used_ts,
+                metrics,
+                results_dict,
+                artifact_version,
+                tf_github_org,
+                tf_github_repo,
+                deployment_name,
+                deployment_type,
+                test_name,
+                tf_triggering_env,
+                metadata_tags,
+                build_variant_name,
+                running_platform,
+                testcase_metric_context_paths,
+            )
+            total_break_by_added += 1
     else:
         logging.warning(
             "there was no git VERSION information to push data brokedown by VERSION"
@@ -1054,6 +1067,9 @@ def add_standardized_metric_bybranch(
         labels["deployment_name+branch"] = "{} {}".format(
             deployment_name, tf_github_branch
         )
+        labels["running_platform+branch"] = "{} {}".format(
+            running_platform, tf_github_branch
+        )
         labels["test_name"] = str(test_name)
         labels["metric"] = str(metric_name)
         logging.info(
@@ -1118,11 +1134,15 @@ def add_standardized_metric_byversion(
             tf_triggering_env,
             metadata_tags,
             build_variant_name,
+            running_platform,
         )
         labels["version"] = artifact_version
         labels["deployment_name+version"] = "{} {}".format(
             deployment_name, artifact_version
         )
+        labels["running_platform+version"] = "{} {}".format(
+            running_platform, artifact_version
+        )
         labels["test_name"] = str(test_name)
         labels["metric"] = str(metric_name)
         logging.info(
@@ -1169,6 +1189,7 @@ def timeseries_test_sucess_flow(
     running_platform=None,
     timeseries_dict=None,
     git_hash=None,
+    disable_target_tables=False,
 ):
     testcase_metric_context_paths = []
     version_target_tables = None
@@ -1205,7 +1226,7 @@ def timeseries_test_sucess_flow(
             )
         )
         push_data_to_redistimeseries(rts, timeseries_dict)
-        if version_target_tables is not None:
+        if not disable_target_tables and version_target_tables is not None:
             logging.info(
                 "There are a total of {} distinct target tables by version".format(
                     len(version_target_tables.keys())
@@ -1225,7 +1246,12 @@ def timeseries_test_sucess_flow(
                 rts.hset(
                     version_target_table_keyname, None, None, version_target_table_dict
                 )
-        if branch_target_tables is not None:
+        elif disable_target_tables:
+            logging.info(
+                "Target tables disabled - skipping version target table creation"
+            )
+
+        if not disable_target_tables and branch_target_tables is not None:
             logging.info(
                 "There are a total of {} distinct target tables by branch".format(
                     len(branch_target_tables.keys())
@@ -1246,6 +1272,10 @@ def timeseries_test_sucess_flow(
                 rts.hset(
                     branch_target_table_keyname, None, None, branch_target_table_dict
                 )
+        elif disable_target_tables:
+            logging.info(
+                "Target tables disabled - skipping branch target table creation"
+            )
         if test_name is not None:
             if type(test_name) is str:
                 update_secondary_result_keys(

redis_benchmarks_specification/__common__/suppress_warnings.py

@@ -46,6 +46,30 @@ def create_compare_arguments(parser):
         default="",
         help="specify a test (or a comma separated list of tests) to use for comparison. If none is specified by default will use all of them.",
     )
+    parser.add_argument(
+        "--extra-filters",
+        type=str,
+        default="",
+        help="specify extra filters to pass to baseline and comparison.",
+    )
+    parser.add_argument(
+        "--use-test-suites-folder",
+        action="store_true",
+        default=False,
+        help="Use test names from YAML files in test-suites folder instead of database",
+    )
+    parser.add_argument(
+        "--generate-boxplot",
+        action="store_true",
+        default=False,
+        help="Generate box plot showing performance change distribution per command",
+    )
+    parser.add_argument(
+        "--boxplot-output",
+        type=str,
+        default="command_performance_boxplot.png",
+        help="Output filename for the box plot (supports .png, .svg, .pdf)",
+    )
     parser.add_argument(
         "--defaults_filename",
         type=str,
@@ -155,6 +179,20 @@ def create_compare_arguments(parser):
     parser.add_argument("--simple-table", type=bool, default=False)
     parser.add_argument("--use_metric_context_path", type=bool, default=False)
     parser.add_argument("--testname_regex", type=str, default=".*", required=False)
+    parser.add_argument(
+        "--command-group-regex",
+        type=str,
+        default=".*",
+        required=False,
+        help="Filter commands by command group using regex. Only commands belonging to matching groups will be included in boxplot and summary.",
+    )
+    parser.add_argument(
+        "--command-regex",
+        type=str,
+        default=".*",
+        required=False,
+        help="Filter tests by command using regex. Only tests that include commands matching this regex will be processed.",
+    )
     parser.add_argument(
         "--regression_str", type=str, default="REGRESSION", required=False
     )