|
| 1 | +import argparse |
| 2 | +import subprocess |
| 3 | +import pathlib |
| 4 | +import os |
| 5 | +from sys import stderr |
| 6 | + |
| 7 | + |
| 8 | +def variant(string): |
| 9 | + if string not in ["h", "ds"]: |
| 10 | + raise ValueError("variant must be h or ds") |
| 11 | + return string |
| 12 | + |
| 13 | + |
| 14 | +def paths(string): |
| 15 | + return list(map(pathlib.Path, string.split(";"))) |
| 16 | + |
| 17 | + |
| 18 | +def parse_args(): |
| 19 | + subparser = argparse.ArgumentParser() |
| 20 | + |
| 21 | + subparser.add_argument('--is-test', action="store_true", default=False) |
| 22 | + |
| 23 | + subparser.add_argument('--datasize', type=int, default=1) |
| 24 | + subparser.add_argument('--variant', type=variant, default='h') |
| 25 | + subparser.add_argument('--tasks', type=int, default=1) |
| 26 | + |
| 27 | + subparser.add_argument('-o', '--output', default="./results") |
| 28 | + subparser.add_argument('--clean-old', action="store_true", default=False) |
| 29 | + subparser.add_argument('--query-filter', action="append", default=[]) |
| 30 | + |
| 31 | + args, argv = subparser.parse_known_args() |
| 32 | + |
| 33 | + if args.is_test: |
| 34 | + parser = argparse.ArgumentParser() |
| 35 | + |
| 36 | + parser.add_argument('--dqrun', type=pathlib.Path) |
| 37 | + parser.add_argument('--gen-queries', type=pathlib.Path) |
| 38 | + parser.add_argument('--downloaders-dir', type=pathlib.Path) |
| 39 | + parser.add_argument('--udfs-dir', type=paths) |
| 40 | + parser.add_argument('--fs-cfg', type=pathlib.Path) |
| 41 | + parser.add_argument('--flame-graph', type=pathlib.Path) |
| 42 | + parser.add_argument('--result-compare', type=pathlib.Path) |
| 43 | + parser.add_argument('--gateways-cfg', type=pathlib.Path) |
| 44 | + parser.add_argument('--runner-path', type=pathlib.Path) |
| 45 | + |
| 46 | + return parser.parse_args(argv, namespace=args) |
| 47 | + else: |
| 48 | + parser = argparse.ArgumentParser() |
| 49 | + |
| 50 | + parser.add_argument('--ydb-root', type=lambda path: pathlib.Path(path).resolve(), default="../../../../") |
| 51 | + |
| 52 | + args = parser.parse_args(argv, namespace=args) |
| 53 | + |
| 54 | + args.dqrun = args.ydb_root / "ydb" / "library" / "yql" / "tools" / "dqrun" / "dqrun" |
| 55 | + args.gen_queries = args.ydb_root / "ydb" / "library" / "benchmarks" / "gen_queries" / "gen_queries" |
| 56 | + args.downloaders_dir = args.ydb_root / "ydb" / "library" / "benchmarks" / "runner" |
| 57 | + args.fs_cfg = args.ydb_root / "ydb" / "library" / "yql" / "tools" / "dqrun" / "examples" / "fs.conf" |
| 58 | + args.flame_graph = args.ydb_root / "contrib" / "tools" / "flame-graph" |
| 59 | + args.result_compare = args.ydb_root / "ydb" / "library" / "benchmarks" / "runner" / "result_compare" / "result_compare" |
| 60 | + args.gateways_cfg = args.ydb_root / "ydb" / "library" / "benchmarks" / "runner" / "runner" / "test-gateways.conf" |
| 61 | + args.runner_path = args.ydb_root / "ydb" / "library" / "benchmarks" / "runner" / "runner" / "runner" |
| 62 | + |
| 63 | + udfs_prefix = args.ydb_root / "ydb" / "library" / "yql" / "udfs" / "common" |
| 64 | + args.udfs_dir = [udfs_prefix / name for name in ["set", "url_base", "datetime2", "re2", "math", "unicode_base"]] |
| 65 | + |
| 66 | + return args |
| 67 | + |
| 68 | + |
| 69 | +class Runner: |
| 70 | + def prepare_queries_dir(self, custom_pragmas): |
| 71 | + print("Preparing queries...", file=stderr) |
| 72 | + self.queries_dir.mkdir(parents=True, exist_ok=True) |
| 73 | + cmd = [str(self.args.gen_queries)] |
| 74 | + cmd += ["--output", f"{self.queries_dir}"] |
| 75 | + cmd += ["--variant", f"{self.args.variant}"] |
| 76 | + cmd += ["--syntax", "yql"] |
| 77 | + cmd += ["--dataset-size", f"{self.args.datasize}"] |
| 78 | + for it in custom_pragmas: |
| 79 | + cmd += ["--pragma", it] |
| 80 | + res = subprocess.run(cmd) |
| 81 | + if res.returncode != 0: |
| 82 | + raise OSError("Failed to prepare queries") |
| 83 | + |
| 84 | + def prepare_tpc_dir(self): |
| 85 | + print("Preparing tpc...", file=stderr) |
| 86 | + cmd = [f"./download_files_{self.args.variant}_{self.args.datasize}.sh"] |
| 87 | + res = subprocess.run(cmd, cwd=self.args.downloaders_dir) |
| 88 | + if res.returncode != 0: |
| 89 | + raise OSError("Failed to prepare tpc") |
| 90 | + |
| 91 | + def __init__(self, args, enable_spilling): |
| 92 | + self.args = args |
| 93 | + self.enable_spilling = enable_spilling |
| 94 | + |
| 95 | + self.queries_dir = pathlib.Path(f"queries{"+" if self.enable_spilling else "-"}spilling-{args.datasize}-{args.tasks}") |
| 96 | + if self.args.clean_old or not self.queries_dir.exists(): |
| 97 | + self.prepare_queries_dir([ |
| 98 | + f"dq.MaxTasksPerStage={self.args.tasks}", |
| 99 | + "dq.OptLLVM=ON" |
| 100 | + ] + [ |
| 101 | + "dq.UseFinalizeByKey=true", |
| 102 | + "dq.EnableSpillingNodes=All", |
| 103 | + ] if self.enable_spilling else []) |
| 104 | + self.tpc_dir = pathlib.Path(f"{self.args.downloaders_dir}/tpc/{self.args.variant}/{self.args.datasize}") |
| 105 | + if self.args.clean_old or not self.tpc_dir.exists(): |
| 106 | + self.prepare_tpc_dir() |
| 107 | + if not pathlib.Path("./tpc").exists(): |
| 108 | + os.symlink(f"{self.args.downloaders_dir}/tpc", f"{pathlib.Path("./tpc")}", target_is_directory=True) |
| 109 | + |
| 110 | + self.result_dir = pathlib.Path(f"{self.args.output}/{"with" if self.enable_spilling else "no"}-spilling/{args.variant}-{args.datasize}-{args.tasks}").resolve() |
| 111 | + self.result_dir.mkdir(parents=True, exist_ok=True) |
| 112 | + |
| 113 | + def run(self): |
| 114 | + cmd = ["/usr/bin/time", f"{str(self.args.runner_path)}"] |
| 115 | + # cmd += ["--perf"] |
| 116 | + for it in self.args.query_filter: |
| 117 | + cmd += ["--include-q", it] |
| 118 | + cmd += ["--query-dir", f"{str(self.queries_dir)}/{self.args.variant}"] |
| 119 | + cmd += ["--bindings", f"{str(self.queries_dir)}/{self.args.variant}/bindings.json"] |
| 120 | + cmd += ["--result-dir", str(self.result_dir)] |
| 121 | + cmd += ["--flame-graph", str(self.args.flame_graph)] |
| 122 | + cmd += [f"{self.args.dqrun}", "-s"] |
| 123 | + cmd += ["--enable-spilling"] if self.enable_spilling else [] |
| 124 | + cmd += ["--udfs-dir", ";".join(map(str, self.args.udfs_dir))] |
| 125 | + cmd += ["--fs-cfg", f"{str(self.args.fs_cfg)}"] |
| 126 | + cmd += ["--gateways-cfg", f"{str(self.args.gateways_cfg)}"] |
| 127 | + print("Running runner...", file=stderr) |
| 128 | + subprocess.run(cmd) |
| 129 | + |
| 130 | + print("Run results at: ", self.result_dir) |
| 131 | + return self.result_dir |
| 132 | + |
| 133 | + |
| 134 | +def result_compare(args, to_compare): |
| 135 | + print("Comparing...") |
| 136 | + cmd = [f"{args.result_compare}"] |
| 137 | + cmd += ["-v"] |
| 138 | + cmd += to_compare |
| 139 | + with open(f"{args.output}/result-{args.variant}-{args.datasize}-{args.tasks}.htm", "w") as result_table: |
| 140 | + res = subprocess.run(cmd, stdout=result_table) |
| 141 | + if res.returncode != 0: |
| 142 | + raise OSError("Failed to compare result") |
| 143 | + |
| 144 | + |
| 145 | +def main(): |
| 146 | + args = parse_args() |
| 147 | + |
| 148 | + results = [] |
| 149 | + print("With spilling...", file=stderr) |
| 150 | + results.append(Runner(args, True).run()) |
| 151 | + print("No spilling...", file=stderr) |
| 152 | + results.append(Runner(args, False).run()) |
| 153 | + |
| 154 | + if not args.is_test: |
| 155 | + result_compare(args, results) |
| 156 | + |
| 157 | + |
| 158 | +if __name__ == "__main__": |
| 159 | + main() |
0 commit comments