ClickHouse/tests/ci/stress_check.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

246 lines
7.4 KiB
Python
Raw Normal View History

2021-11-01 10:27:46 +00:00
#!/usr/bin/env python3
import csv
import logging
import subprocess
import os
import sys
from pathlib import Path
2022-11-16 09:24:10 +00:00
from typing import List, Tuple
2021-11-01 10:27:46 +00:00
from github import Github
2021-11-12 11:07:54 +00:00
from build_download_helper import download_all_deb_packages
2022-03-07 13:25:06 +00:00
from clickhouse_helper import (
ClickHouseHelper,
prepare_tests_results_for_clickhouse,
2023-08-12 20:30:18 +00:00
get_instance_type,
2022-03-07 13:25:06 +00:00
)
from commit_status_helper import RerunHelper, get_commit, post_commit_status
from docker_pull_helper import get_image_with_version
from env_helper import TEMP_PATH, REPO_COPY, REPORTS_PATH
from get_robot_token import get_best_robot_token
from pr_info import PRInfo
from report import TestResults, read_test_results
from s3_helper import S3Helper
from stopwatch import Stopwatch
2021-12-03 08:33:16 +00:00
from tee_popen import TeePopen
from upload_result_helper import upload_results
2021-11-19 14:47:04 +00:00
2021-11-01 10:27:46 +00:00
2022-03-07 13:25:06 +00:00
def get_run_command(
2023-08-12 20:30:18 +00:00
pr_info,
check_start_time,
check_name,
build_path,
result_folder,
repo_tests_path,
server_log_folder,
2023-08-12 20:43:29 +00:00
image,
2022-03-07 13:25:06 +00:00
):
2023-08-12 20:30:18 +00:00
instance_type = get_instance_type()
envs = [
# a static link, don't use S3_URL or S3_DOWNLOAD
"-e S3_URL='https://s3.amazonaws.com/clickhouse-datasets'",
"-e CLICKHOUSE_CI_LOGS_HOST",
"-e CLICKHOUSE_CI_LOGS_PASSWORD",
f"-e PULL_REQUEST_NUMBER='{pr_info.number}'",
f"-e COMMIT_SHA='{pr_info.sha}'",
f"-e CHECK_START_TIME='{check_start_time}'",
f"-e CHECK_NAME='{check_name}'",
f"-e INSTANCE_TYPE='{instance_type}'",
]
env_str = " ".join(envs)
2022-03-07 13:25:06 +00:00
cmd = (
"docker run --cap-add=SYS_PTRACE "
2023-08-12 20:30:18 +00:00
f"{env_str} "
# For dmesg and sysctl
"--privileged "
2022-03-07 13:25:06 +00:00
f"--volume={build_path}:/package_folder "
f"--volume={result_folder}:/test_output "
f"--volume={repo_tests_path}:/usr/share/clickhouse-test "
2022-06-01 08:44:44 +00:00
f"--volume={server_log_folder}:/var/log/clickhouse-server {image} "
2022-03-07 13:25:06 +00:00
)
2021-11-01 10:27:46 +00:00
return cmd
2022-03-07 13:25:06 +00:00
2022-11-16 09:24:10 +00:00
def process_results(
result_folder: str, server_log_path: str, run_log_path: str
) -> Tuple[str, str, TestResults, List[str]]:
test_results = [] # type: TestResults
2021-11-01 10:27:46 +00:00
additional_files = []
# Just upload all files from result_folder.
2022-03-07 13:25:06 +00:00
# If task provides processed results, then it's responsible for content
# of result_folder.
2021-11-01 10:27:46 +00:00
if os.path.exists(result_folder):
2022-03-07 13:25:06 +00:00
test_files = [
f
for f in os.listdir(result_folder)
if os.path.isfile(os.path.join(result_folder, f))
]
2021-11-01 10:27:46 +00:00
additional_files = [os.path.join(result_folder, f) for f in test_files]
if os.path.exists(server_log_path):
2022-03-07 13:25:06 +00:00
server_log_files = [
f
for f in os.listdir(server_log_path)
if os.path.isfile(os.path.join(server_log_path, f))
]
additional_files = additional_files + [
os.path.join(server_log_path, f) for f in server_log_files
]
2021-11-01 10:27:46 +00:00
2021-11-01 13:27:55 +00:00
additional_files.append(run_log_path)
2021-11-01 10:27:46 +00:00
status_path = os.path.join(result_folder, "check_status.tsv")
if not os.path.exists(status_path):
2022-03-07 13:25:06 +00:00
return (
"failure",
"check_status.tsv doesn't exists",
test_results,
additional_files,
)
2021-11-01 10:27:46 +00:00
logging.info("Found check_status.tsv")
2022-03-07 13:25:06 +00:00
with open(status_path, "r", encoding="utf-8") as status_file:
status = list(csv.reader(status_file, delimiter="\t"))
2021-11-01 10:27:46 +00:00
if len(status) != 1 or len(status[0]) != 2:
return "error", "Invalid check_status.tsv", test_results, additional_files
state, description = status[0][0], status[0][1]
try:
results_path = Path(result_folder) / "test_results.tsv"
test_results = read_test_results(results_path, True)
if len(test_results) == 0:
raise Exception("Empty results")
except Exception as e:
return (
"error",
f"Cannot parse test_results.tsv ({e})",
test_results,
additional_files,
)
2021-11-01 10:27:46 +00:00
return state, description, test_results, additional_files
def run_stress_test(docker_image_name):
2021-11-01 10:27:46 +00:00
logging.basicConfig(level=logging.INFO)
2021-11-19 14:47:04 +00:00
stopwatch = Stopwatch()
2021-11-26 14:00:09 +00:00
temp_path = TEMP_PATH
repo_path = REPO_COPY
2022-02-15 12:03:51 +00:00
repo_tests_path = os.path.join(repo_path, "tests")
2021-11-26 14:00:09 +00:00
reports_path = REPORTS_PATH
2021-11-01 10:27:46 +00:00
check_name = sys.argv[1]
if not os.path.exists(temp_path):
os.makedirs(temp_path)
2021-11-26 14:00:09 +00:00
pr_info = PRInfo()
2021-11-01 10:27:46 +00:00
gh = Github(get_best_robot_token(), per_page=100)
commit = get_commit(gh, pr_info.sha)
2021-11-01 10:27:46 +00:00
rerun_helper = RerunHelper(commit, check_name)
2021-12-01 14:23:51 +00:00
if rerun_helper.is_already_finished_by_status():
logging.info("Check is already finished according to github status, exiting")
sys.exit(0)
docker_image = get_image_with_version(reports_path, docker_image_name)
2021-11-01 10:27:46 +00:00
packages_path = os.path.join(temp_path, "packages")
if not os.path.exists(packages_path):
os.makedirs(packages_path)
2021-11-12 11:07:54 +00:00
download_all_deb_packages(check_name, reports_path, packages_path)
2021-11-01 10:27:46 +00:00
server_log_path = os.path.join(temp_path, "server_log")
if not os.path.exists(server_log_path):
os.makedirs(server_log_path)
result_path = os.path.join(temp_path, "result_path")
if not os.path.exists(result_path):
os.makedirs(result_path)
2022-12-26 15:29:32 +00:00
run_log_path = os.path.join(temp_path, "run.log")
2021-11-01 10:27:46 +00:00
2022-03-07 13:25:06 +00:00
run_command = get_run_command(
2023-08-12 20:30:18 +00:00
pr_info,
stopwatch.start_time_str,
check_name,
packages_path,
result_path,
repo_tests_path,
server_log_path,
2023-08-12 20:43:29 +00:00
docker_image,
2022-03-07 13:25:06 +00:00
)
2023-08-12 20:30:18 +00:00
logging.info("Going to run stress test: %s", run_command)
2021-11-01 10:27:46 +00:00
2022-12-14 10:53:52 +00:00
with TeePopen(run_command, run_log_path, timeout=60 * 150) as process:
2021-12-03 08:33:16 +00:00
retcode = process.wait()
if retcode == 0:
logging.info("Run successfully")
else:
logging.info("Run failed")
2021-11-01 10:27:46 +00:00
subprocess.check_call(f"sudo chown -R ubuntu:ubuntu {temp_path}", shell=True)
2022-08-11 13:01:32 +00:00
s3_helper = S3Helper()
2022-03-07 13:25:06 +00:00
state, description, test_results, additional_logs = process_results(
result_path, server_log_path, run_log_path
)
2021-11-19 14:47:04 +00:00
ch_helper = ClickHouseHelper()
2023-08-12 20:30:18 +00:00
# Cleanup run log from the credentials of CI logs database.
# Note: a malicious user can still print them by splitting the value into parts.
# But we will be warned when a malicious user modifies CI script.
# Although they can also print them from inside tests.
# Nevertheless, the credentials of the CI logs have limited scope
# and does not provide access to sensitive info.
ci_logs_host = os.getenv("CLICKHOUSE_CI_LOGS_HOST", "CLICKHOUSE_CI_LOGS_HOST")
ci_logs_password = os.getenv(
"CLICKHOUSE_CI_LOGS_PASSWORD", "CLICKHOUSE_CI_LOGS_PASSWORD"
)
subprocess.check_call(
f"sed -i -r -e 's!{ci_logs_host}!CLICKHOUSE_CI_LOGS_HOST!g; s!{ci_logs_password}!CLICKHOUSE_CI_LOGS_PASSWORD!g;' '{run_log_path}'",
shell=True,
)
2022-03-07 13:25:06 +00:00
report_url = upload_results(
s3_helper,
pr_info.number,
pr_info.sha,
test_results,
2022-03-07 13:31:17 +00:00
additional_logs,
2022-03-07 13:25:06 +00:00
check_name,
)
2021-11-01 10:27:46 +00:00
print(f"::notice ::Report url: {report_url}")
2021-11-12 12:36:25 +00:00
post_commit_status(commit, state, report_url, description, check_name, pr_info)
2021-11-19 14:47:04 +00:00
2022-03-07 13:25:06 +00:00
prepared_events = prepare_tests_results_for_clickhouse(
pr_info,
test_results,
state,
stopwatch.duration_seconds,
stopwatch.start_time_str,
report_url,
check_name,
)
2022-03-29 19:06:50 +00:00
ch_helper.insert_events_into(db="default", table="checks", events=prepared_events)
2022-03-29 12:41:47 +00:00
if state == "failure":
2022-03-29 12:41:47 +00:00
sys.exit(1)
2023-01-03 19:32:11 +00:00
if __name__ == "__main__":
run_stress_test("clickhouse/stress-test")