Updated the benchmark SPEC to include command priority. Included --dry-run and --tests-priority-upper-limit in redis-benchmarks-spec-client-runner (#184)

Author: filipecosta90

commands-priority.json

@@ -74,84 +74,84 @@
  "setbit": 73,
  "zcount": 74,
  "unlink": 75,
- "select": 76,
- "setnx": 77,
- "type": 78,
- "hlen": 79,
- "sscan": 80,
- "hsetnx": 81,
+ "setnx": 76,
+ "type": 77,
+ "hlen": 78,
+ "sscan": 79,
+ "hsetnx": 80,
+ "select": 81,
  "srandmember": 82,
  "zscan": 83,
  "xack": 84,
  "pttl": 85,
  "restore": 86,
  "pexpireat": 87,
  "xdel": 88,
- "discard": 89,
- "xpending": 90,
- "getset": 91,
- "psetex": 92,
- "mset": 93,
- "bzpopmin": 94,
- "persist": 95,
- "pfcount": 96,
- "hkeys": 97,
- "xread": 98,
- "xlen": 99,
- "xtrim": 100,
- "decrby": 101,
- "rpushx": 102,
- "sinter": 103,
- "decr": 104,
- "sunion": 105,
- "sinterstore": 106,
- "smismember": 107,
- "zpopmin": 108,
- "echo": 109,
- "dbsize": 110,
- "lset": 111,
+ "xpending": 89,
+ "getset": 90,
+ "psetex": 91,
+ "mset": 92,
+ "bzpopmin": 93,
+ "persist": 94,
+ "pfcount": 95,
+ "hkeys": 96,
+ "xread": 97,
+ "xlen": 98,
+ "xtrim": 99,
+ "decrby": 100,
+ "rpushx": 101,
+ "sinter": 102,
+ "decr": 103,
+ "sunion": 104,
+ "sinterstore": 105,
+ "smismember": 106,
+ "zpopmin": 107,
+ "echo": 108,
+ "dbsize": 109,
+ "lset": 110,
+ "geoadd": 111,
  "xclaim": 112,
- "geoadd": 113,
- "randomkey": 114,
- "rename": 115,
- "zrangebylex": 116,
- "zinterstore": 117,
- "append": 118,
- "zunionstore": 119,
- "xrange": 120,
- "getrange": 121,
- "xautoclaim": 122,
- "sunionstore": 123,
- "bitcount": 124,
- "dump": 125,
- "hrandfield": 126,
- "getex": 127,
- "strlen": 128,
- "pubsub": 129,
- "linsert": 130,
- "bitpos": 131,
- "sdiff": 132,
- "msetnx": 133,
- "getdel": 134,
- "sort": 135,
- "renamenx": 136,
- "setrange": 137,
- "lpos": 138,
- "geopos": 139,
- "blmove": 140,
- "georadius": 141,
- "zlexcount": 142,
- "pfmerge": 143,
- "xrevrange": 144,
- "georadiusbymember": 145,
- "zremrangebylex": 146,
- "lmove": 147,
- "sdiffstore": 148,
- "zpopmax": 149,
- "bzpopmax": 150,
- "zmscore": 151,
- "flushdb": 152,
- "zrevrangebylex": 153,
+ "randomkey": 113,
+ "rename": 114,
+ "zrangebylex": 115,
+ "zinterstore": 116,
+ "append": 117,
+ "zunionstore": 118,
+ "xrange": 119,
+ "getrange": 120,
+ "xautoclaim": 121,
+ "sunionstore": 122,
+ "bitcount": 123,
+ "dump": 124,
+ "hrandfield": 125,
+ "getex": 126,
+ "strlen": 127,
+ "pubsub": 128,
+ "linsert": 129,
+ "bitpos": 130,
+ "sdiff": 131,
+ "msetnx": 132,
+ "getdel": 133,
+ "sort": 134,
+ "renamenx": 135,
+ "setrange": 136,
+ "lpos": 137,
+ "geopos": 138,
+ "blmove": 139,
+ "georadius": 140,
+ "zlexcount": 141,
+ "pfmerge": 142,
+ "xrevrange": 143,
+ "georadiusbymember": 144,
+ "zremrangebylex": 145,
+ "lmove": 146,
+ "sdiffstore": 147,
+ "zpopmax": 148,
+ "bzpopmax": 149,
+ "zmscore": 150,
+ "flushdb": 151,
+ "zrevrangebylex": 152,
+ "discard": 153,
  "smove": 154,
  "flushall": 155,
  "georadius_ro": 156,

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "redis-benchmarks-specification"
-version = "0.1.61"
+version = "0.1.62"
 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/__cli__/stats.py

@@ -44,6 +44,15 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
             len(testsuite_spec_files), testsuites_folder
         )
     )
+    priority_json = {}
+    if args.commands_priority_file != "":
+        with open(args.commands_priority_file, "r") as fd:
+            logging.info(
+                "Reading {} file with priority by commandstats".format(
+                    args.commands_priority_file
+                )
+            )
+            priority_json = json.load(fd)
     tracked_groups = []
     override_enabled = args.override_tests
     fail_on_required_diff = args.fail_on_required_diff
@@ -76,6 +85,7 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
                 if "tested-groups" in benchmark_config:
                     origin_tested_groups = benchmark_config["tested-groups"]
                 origin_tested_commands = []
+
                 tested_commands = []
                 if "tested-commands" in benchmark_config:
                     origin_tested_commands = benchmark_config["tested-commands"]
@@ -126,6 +136,31 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
                         )
                     )
 
+                priority = None
+                # maximum priority of all tested commands
+                priority_json_value = None
+                for command in tested_commands:
+                    if command in priority_json:
+                        priority_v = priority_json[command]
+                        if priority_json_value is None:
+                            priority_json_value = priority_v
+                        if priority_v > priority_json_value:
+                            priority_json_value = priority_v
+
+                if "priority" in benchmark_config:
+                    priority = benchmark_config["priority"]
+                else:
+                    if priority_json_value is not None:
+                        requires_override = True
+                        logging.warn(
+                            "dont have priority in {}, but the commands in the test have max priority of {}".format(
+                                test_name, priority_json_value
+                            )
+                        )
+                        priority = priority_json_value
+                if priority is not None:
+                    benchmark_config["priority"] = priority
+
                 if tested_groups != origin_tested_groups:
                     requires_override = True
                     benchmark_config["tested-groups"] = tested_groups
@@ -171,6 +206,7 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
 
         # open file in read mode
         total_count = 0
+        total_usecs = 0
         total_tracked_count = 0
         with open(
             args.commandstats_csv, "r", encoding="utf8", errors="ignore"
@@ -185,6 +221,10 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
                 cmdstat = row[0]
                 cmdstat = cmdstat.replace("cmdstat_", "")
                 count = int(row[1])
+                usecs = None
+                if len(row) > 2:
+                    usecs = int(row[2])
+                    total_usecs += usecs
                 if count == 0:
                     continue
                 tracked = False
@@ -206,7 +246,7 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
                 if cmdstat in tracked_commands_json:
                     tracked = True
                 if module is False or include_modules:
-                    row = [cmdstat, group, count, tracked, deprecated]
+                    row = [cmdstat, group, count, usecs, tracked, deprecated]
                     rows.append(row)
 
         priority_list = sorted(((priority[cmd], cmd) for cmd in priority), reverse=True)
@@ -245,7 +285,16 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
                 json.dump(priority_json, fd, indent=True)
 
         if args.summary_csv != "":
-            header = ["command", "group", "count", "tracked", "deprecated", "%"]
+            header = [
+                "command",
+                "group",
+                "count",
+                "usecs",
+                "tracked",
+                "deprecated",
+                "% count",
+                "% usecs",
+            ]
             import csv
 
             with open(args.summary_csv, "w", encoding="UTF8", newline="") as f:
@@ -256,8 +305,13 @@ def generate_stats_cli_command_logic(args, project_name, project_version):
                 for row in rows:
                     # write the data
                     count = row[2]
+                    usec = row[3]
                     pct = count / total_count
+                    pct_usec = "n/a"
+                    if usec is not None:
+                        pct_usec = usec / total_usecs
                     row.append(pct)
+                    row.append(pct_usec)
                     writer.writerow(row)
 
         if total_tracked_count > 0:

redis_benchmarks_specification/__runner__/args.py

@@ -56,6 +56,18 @@ def create_client_runner_args(project_name):
     parser.add_argument("--db_server_host", type=str, default="localhost")
     parser.add_argument("--db_server_port", type=int, default=6379)
     parser.add_argument("--cpuset_start_pos", type=int, default=0)
+    parser.add_argument(
+        "--tests-priority-upper-limit",
+        type=int,
+        default=-1,
+        help="Run a subset of the tests based uppon a preset priority. By default runs all tests.",
+    )
+    parser.add_argument(
+        "--dry-run",
+        default=False,
+        action="store_true",
+        help="Only check how many benchmarks we would run. Don't run benchmark but can change state of DB.",
+    )
     parser.add_argument(
         "--datasink_redistimeseries_host", type=str, default=DATASINK_RTS_HOST
     )

redis_benchmarks_specification/__runner__/runner.py

@@ -262,6 +262,8 @@ def process_self_contained_coordinator_stream(
     overall_result = True
     results_matrix = []
     total_test_suite_runs = 0
+    dry_run_count = 0
+    dry_run = args.dry_run
     for test_file in testsuite_spec_files:
         client_containers = []
 
@@ -290,6 +292,7 @@ def process_self_contained_coordinator_stream(
                     tf_github_repo = args.github_repo
                     tf_triggering_env = args.platform_name
                     setup_type = args.setup_type
+                    priority_limit = args.tests_priority_upper_limit
                     git_hash = "NA"
                     git_version = args.github_version
                     build_variant_name = "NA"
@@ -364,6 +367,25 @@ def process_self_contained_coordinator_stream(
                             _, test_tls_key = cp_to_workdir(
                                 temporary_dir_client, tls_key
                             )
+                    priority = None
+                    if "priority" in benchmark_config:
+                        priority = benchmark_config["priority"]
+
+                        if priority_limit > 0 and priority is not None:
+                            if priority_limit < priority:
+                                logging.warning(
+                                    "Skipping test {} giving the priority limit ({}) is above the priority value ({})".format(
+                                        test_name, priority_limit, priority
+                                    )
+                                )
+                                continue
+                            logging.info(
+                                "Test {} priority ({}) is within the priority limit ({})".format(
+                                    test_name,
+                                    priority,
+                                    priority_limit,
+                                )
+                            )
 
                     if "dataset" in benchmark_config["dbconfig"]:
                         if args.run_tests_with_dataset is False:
@@ -374,6 +396,10 @@ def process_self_contained_coordinator_stream(
                             )
                             continue
 
+                    if dry_run is True:
+                        dry_run_count = dry_run_count + 1
+                        continue
+
                     if "preload_tool" in benchmark_config["dbconfig"]:
                         data_prepopulation_step(
                             benchmark_config,
@@ -669,6 +695,11 @@ def process_self_contained_coordinator_stream(
         )
         csv_writer.dump(dest_fpath)
 
+    if dry_run is True:
+        logging.info(
+            "Number of tests that would have been run: {}".format(dry_run_count)
+        )
+
 
 def cp_to_workdir(benchmark_tool_workdir, srcfile):
     head, filename = os.path.split(srcfile)

redis_benchmarks_specification/test-suites/memtier_benchmark-10Mkeys-load-hash-5-fields-with-1000B-values-pipeline-10.yml

@@ -31,3 +31,4 @@ exporter:
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
 tested-groups:
 - hash
+priority: 5

redis_benchmarks_specification/test-suites/memtier_benchmark-10Mkeys-load-hash-5-fields-with-1000B-values.yml

@@ -31,3 +31,4 @@ exporter:
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
 tested-groups:
 - hash
+priority: 5

redis_benchmarks_specification/test-suites/memtier_benchmark-10Mkeys-load-hash-5-fields-with-100B-values-pipeline-10.yml

@@ -31,3 +31,4 @@ exporter:
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
 tested-groups:
 - hash
+priority: 5

redis_benchmarks_specification/test-suites/memtier_benchmark-10Mkeys-load-hash-5-fields-with-100B-values.yml

@@ -31,3 +31,4 @@ exporter:
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
 tested-groups:
 - hash
+priority: 5

redis_benchmarks_specification/test-suites/memtier_benchmark-10Mkeys-load-hash-5-fields-with-10B-values-pipeline-10.yml

@@ -31,3 +31,4 @@ exporter:
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
 tested-groups:
 - hash
+priority: 5

redis_benchmarks_specification/test-suites/memtier_benchmark-10Mkeys-load-hash-5-fields-with-10B-values.yml

@@ -31,3 +31,4 @@ exporter:
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
 tested-groups:
 - hash
+priority: 5

redis_benchmarks_specification/test-suites/memtier_benchmark-1Mkeys-100B-expire-use-case.yml

@@ -40,3 +40,4 @@ exporter:
     - $."ALL STATS".Totals."Latency"
     - $."ALL STATS".Totals."Misses/sec"
     - $."ALL STATS".Totals."Percentile Latencies"."p50.00"
+priority: 33