import subprocess import os import logging import sys import re g_percentage_msg = None g_has_error = False def run_grep_command(grep_command_obj, filepath): """使用自定义的grep命令从文件中提取数字，并返回数字列表。""" try: grep_command = grep_command_obj["cmd"] # 构建完整的grep命令 command = f"cat {filepath} | {grep_command} " print(command) # 运行命令并捕获输出 result = subprocess.run(command, shell=True, text=True, capture_output=True) # 提取输出中的数字 numbers = list() for line in result.stdout.splitlines(): matches = re.findall(r'[0-9]+(?:\.[0-9]+)?', line) numbers.extend([float(match) for match in matches]) return numbers except Exception as e: print(f"Error running grep command: {e}") return [] def compare_numbers(numbers1, numbers2, threshold=0.05): """比较两个数字列表，检查差异是否都在阈值以内。""" print(numbers1, numbers2) if len(numbers1) != len(numbers2): return False for num1, num2 in zip(numbers1, numbers2): if max(num1, num2) == 0: if num1 != num2: return False else: continue if abs(num1 - num2) / max(num1, num2) > threshold: print("not within threshold {} {}".format(num1, num2)) percentage = abs(num1 - num2) / max(num1, num2) print(f"percentage: {percentage}") g_percentage_msg = f"偏移: {percentage}" return False return True """比较两个数字列表，检查差异是否大于""" def compare_numbers_greater(numbers1, numbers2): """比较两个数字列表，检查差异是否大于阈值。""" if len(numbers1) != len(numbers2): return False for num1, num2 in zip(numbers1, numbers2): if num1 < num2: print("not greater {} {}".format(num1, num2)) percentage = abs(num1 - num2) / max(num1, num2) print(f"percentage: {percentage}") g_percentage_msg = f"偏移: {percentage}" return False return True def compare_numbers_with_tolerance_greater(numbers1, numbers2, tolerance=0.05): """ 比较两个数字列表，检查第一个列表的元素是否大于第二个列表的对应元素。 如果不大于，但差异在容忍的百分比内，也认为通过。 参数： - numbers1: 第一个数字列表。 - numbers2: 第二个数字列表。 - tolerance: 容忍的百分比。 返回： - True 如果所有比较都通过，否则 False。 """ if len(numbers1) != len(numbers2): return False for num1, num2 in zip(numbers1, numbers2): if num1 < num2: diff = abs(num1 - num2) if num2 != 0 and (diff / num2) > tolerance: print(f"不通过: {num1} < {num2}, 差异百分比: {diff / num2:.2%} 超过容忍值: {tolerance:.2%}") return False else: #print(f"通过: {num1} < {num2}, 但差异百分比: {diff / num2:.2%} 在容忍值: {tolerance:.2%} 内") pass else: #print(f"通过: {num1} >= {num2}") pass return True """比较两个数字列表，检查是否小于""" def compare_numbers_less(numbers1, numbers2): """比较两个数字列表，检查差异是否小于阈值。""" if len(numbers1) != len(numbers2): return False for num1, num2 in zip(numbers1, numbers2): if num1 > num2: print("not less {} {}".format(num1, num2)) percentage = abs(num1 - num2) / max(num1, num2) print(f"percentage: {percentage}") g_percentage_msg = f"偏移: {percentage}" return False return True def compare_numbers_with_tolerance_less(numbers1, numbers2, tolerance=0.05): """ 比较两个数字列表，检查第一个列表的元素是否小于第二个列表的对应元素。 如果不小于，但差异在容忍的百分比内，也认为通过。 参数： - numbers1: 第一个数字列表。 - numbers2: 第二个数字列表。 - tolerance: 容忍的百分比。 返回： - True 如果所有比较都通过，否则 False。 """ if len(numbers1) != len(numbers2): return False for num1, num2 in zip(numbers1, numbers2): if num1 > num2: diff = abs(num1 - num2) if num2 != 0 and (diff / num2) > tolerance: print(f"不通过: {num1} > {num2}, 差异百分比: {diff / num2:.2%} 超过容忍值: {tolerance:.2%}") return False else: #print(f"通过: {num1} > {num2}, 但差异百分比: {diff / num2:.2%} 在容忍值: {tolerance:.2%} 内") pass else: #print(f"通过: {num1} <= {num2}") pass return True def compare_files(log_files, answer_files, grep_commands, result_file): """使用自定义的grep命令比较多组日志文件和答案文件，将结果写入结果文件。""" g_has_error = False common_file_prefix = '/root/presurelog/' ans_file_prefix = '/root/presure-test-tool/targetlog/' with open(result_file, 'w') as result: for log_file, answer_file, grep_command in zip(log_files, answer_files, grep_commands): try: log_numbers = run_grep_command(grep_command, common_file_prefix+log_file) answer_numbers = run_grep_command(grep_command, ans_file_prefix+answer_file) print('start diff %s and %s' % (log_file, answer_file)) if grep_command['comparison'] == 'within_percentage': is_normal = compare_numbers(log_numbers, answer_numbers, threshold=grep_command['percentage']) elif grep_command['comparison'] == 'less': is_normal = compare_numbers_with_tolerance_less(log_numbers, answer_numbers, tolerance=grep_command['percentage']) elif grep_command['comparison'] == 'greater': is_normal = compare_numbers_with_tolerance_greater(log_numbers, answer_numbers, tolerance=grep_command['percentage']) # add log if not is_normal: print(f"{os.path.basename(log_file)}: 异常") print(g_percentage_msg) else: print(f"{os.path.basename(log_file)}: 正常") print(log_numbers) print(answer_numbers) diff_result = "正常" if is_normal else "异常" if diff_result == "异常": g_has_error = True final_str = os.path.basename(log_file)+ " " + diff_result + "\n" result.write(final_str) except Exception as e: logging.error(f"Error comparing {log_file} and {answer_file}: {e}") return g_has_error # 示例文件路径列表和对应的grep命令 analysis_config = { "check_list": [ { "file":"test_all_reduce_perf.log", "cmd": "grep float| grep -Eo '[0-9]+(\.[0-9]+)?' ", "comparison": "greater", "percentage": 0.05 }, #{ # "file":"test_cpu.log", "cmd": "grep total ", "comparison": "greater", "percentage": 0.05 #}, { "file":"test_fio.log", "cmd": "grep IOPS | awk -F'/' '{print $1" "$2}' ", "comparison": "greater", "percentage": 0.05 }, { "file":"test_gpu.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "greater", "percentage": 0.10 }, { "file":"test_mlc_latency_matrix.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "less", "percentage": 0.05 }, { "file":"test_mlc_max_bandwidth.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "greater", "percentage": 0.05 }, { "file":"test_pcie_p2p.log", "cmd": "grep SUM ", "comparison": "greater", "percentage": 0.05 }, ], "result_file": '/root/comparison_pressure_result.txt' } ''' analysis_config = { "check_list": [ #{ # "file":"test_all_reduce_perf.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "greater", "percentage": 0.05 #}, { "file":"test_cpu.log", "cmd": "grep total ", "comparison": "within_percentage", "percentage": 0.1 }, #{ # "file":"test_fio.log", "cmd": "grep IOPS | awk -F'/' '{print $1" "$2}' ", "comparison": "greater", "percentage": 0.05 #}, { "file":"test_gpu.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "within_percentage", "percentage": 0.1 }, { "file":"test_mlc_latency_matrix.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "within_percentage", "percentage": 0.1 }, { "file":"test_mlc_max_bandwidth.log", "cmd": "grep -Eo '[0-9]+(\.[0-9]+)?'", "comparison": "within_percentage", "percentage": 0.1 }, { "file":"test_pcie_p2p.log", "cmd": "grep SUM ", "comparison": "within_percentage", "percentage": 0.1 }, ], "result_file": '/root/comparison_pressure_result.txt' } ''' # 日志文件路径列表 log_files = list() for e_config in analysis_config['check_list']: log_files.append(e_config['file']) # 答案文件路径列表 answer_files = log_files # "grep -Eo '[0-9]+(\.[0-9]+)?'", # 对第一对文件使用的grep命令 # "grep -Eo '[0-9]+'", # 对第二对文件使用的grep命令（仅匹配整数） grep_commands = analysis_config['check_list'] result_file = analysis_config['result_file'] # 运行比较 ret = compare_files(log_files, answer_files, grep_commands, result_file) # run cat /root/comparison_pressure_result.txt command = "cat /root/comparison_pressure_result.txt" result = subprocess.run(command, shell=True, text=True, capture_output=True) print(result.stdout) if ret: print("有异常的压测结果") sys.exit(1)