OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
c0695bb47337870581bdacd635e931b416cea7d4
third_party_mesa3d/.gitlab-ci/lava/lava_job_submitter.py
Guilherme Gallo addac10443 ci: Make LAVA jobs fail CI job when retry is exhausted 
When the lava_job_submitter.py retry loop finishes normally (without
falling through break-loop) it means that the submitter has exceeded the
retry count limit. However, when it happens the script
finishes normally. This patch adds a treatment to this case, warning the
user what happened and forcing the job to fail.

Moreover, this commit will make retry configurations configurable by
CI job, as it can take the default value from the following variables:

- LAVA_DEVICE_HANGING_TIMEOUT_SEC
- LAVA_WAIT_FOR_DEVICE_POLLING_TIME_SEC
- LAVA_LOG_POLLING_TIME_SEC
- LAVA_NUMBER_OF_RETRIES_TIMEOUT_DETECTION

Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14876>
2022-02-16 23:32:39 +00:00

381 lines
13 KiB
Python
Executable File
Raw Blame History

#!/usr/bin/env python3
#
# Copyright (C) 2020, 2021 Collabora Limited
# Author: Gustavo Padovan <gustavo.padovan@collabora.com>
#
# Permission is hereby granted, free of charge, to any person obtaining a
# copy of this software and associated documentation files (the "Software"),
# to deal in the Software without restriction, including without limitation
# the rights to use, copy, modify, merge, publish, distribute, sublicense,
# and/or sell copies of the Software, and to permit persons to whom the
# Software is furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice (including the next
# paragraph) shall be included in all copies or substantial portions of the
# Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
"""Send a job to LAVA, track it and collect log back"""
import argparse
import pathlib
import sys
import time
import traceback
import urllib.parse
import xmlrpc
from datetime import datetime, timedelta
from os import getenv
import lavacli
import yaml
from lavacli.utils import loader
# Timeout in seconds to decide if the device from the dispatched LAVA job has
# hung or not due to the lack of new log output.
DEVICE_HANGING_TIMEOUT_SEC = int(getenv("LAVA_DEVICE_HANGING_TIMEOUT_SEC", 5*60))
# How many seconds the script should wait before try a new polling iteration to
# check if the dispatched LAVA job is running or waiting in the job queue.
WAIT_FOR_DEVICE_POLLING_TIME_SEC = int(getenv("LAVA_WAIT_FOR_DEVICE_POLLING_TIME_SEC", 10))
# How many seconds to wait between log output LAVA RPC calls.
LOG_POLLING_TIME_SEC = int(getenv("LAVA_LOG_POLLING_TIME_SEC", 5))
# How many retries should be made when a timeout happen.
NUMBER_OF_RETRIES_TIMEOUT_DETECTION = int(getenv("LAVA_NUMBER_OF_RETRIES_TIMEOUT_DETECTION", 2))
def print_log(msg):
print("{}: {}".format(datetime.now(), msg))
def fatal_err(msg):
print_log(msg)
sys.exit(1)
def hide_sensitive_data(yaml_data, hide_tag="HIDEME"):
return "".join(line for line in yaml_data.splitlines(True) if hide_tag not in line)
def generate_lava_yaml(args):
# General metadata and permissions, plus also inexplicably kernel arguments
values = {
'job_name': 'mesa: {}'.format(args.pipeline_info),
'device_type': args.device_type,
'visibility': { 'group': [ args.visibility_group ] },
'priority': 75,
'context': {
'extra_nfsroot_args': ' init=/init rootwait minio_results={}'.format(args.job_artifacts_base)
},
'timeouts': {
'job': {
'minutes': args.job_timeout
}
},
}
if args.lava_tags:
values['tags'] = args.lava_tags.split(',')
# URLs to our kernel rootfs to boot from, both generated by the base
# container build
deploy = {
'timeout': { 'minutes': 10 },
'to': 'tftp',
'os': 'oe',
'kernel': {
'url': '{}/{}'.format(args.base_system_url_prefix, args.kernel_image_name),
},
'nfsrootfs': {
'url': '{}/lava-rootfs.tgz'.format(args.base_system_url_prefix),
'compression': 'gz',
}
}
if args.kernel_image_type:
deploy['kernel']['type'] = args.kernel_image_type
if args.dtb:
deploy['dtb'] = {
'url': '{}/{}.dtb'.format(args.base_system_url_prefix, args.dtb)
}
# always boot over NFS
boot = {
'timeout': { 'minutes': 25 },
'method': args.boot_method,
'commands': 'nfs',
'prompts': ['lava-shell:'],
}
# skeleton test definition: only declaring each job as a single 'test'
# since LAVA's test parsing is not useful to us
test = {
'timeout': { 'minutes': args.job_timeout },
'failure_retry': 1,
'definitions': [ {
'name': 'mesa',
'from': 'inline',
'path': 'inline/mesa.yaml',
'repository': {
'metadata': {
'name': 'mesa',
'description': 'Mesa test plan',
'os': [ 'oe' ],
'scope': [ 'functional' ],
'format': 'Lava-Test Test Definition 1.0',
},
'parse': {
'pattern': r'hwci: (?P<test_case_id>\S*):\s+(?P<result>(pass|fail))'
},
'run': {
},
},
} ],
}
# job execution script:
# - inline .gitlab-ci/common/init-stage1.sh
# - fetch and unpack per-pipeline build artifacts from build job
# - fetch and unpack per-job environment from lava-submit.sh
# - exec .gitlab-ci/common/init-stage2.sh
init_lines = []
with open(args.first_stage_init, 'r') as init_sh:
init_lines += [ x.rstrip() for x in init_sh if not x.startswith('#') and x.rstrip() ]
with open(args.jwt_file) as jwt_file:
init_lines += [
"set +x",
f'echo -n "{jwt_file.read()}" > "{args.jwt_file}" # HIDEME',
"set -x",
]
init_lines += [
'mkdir -p {}'.format(args.ci_project_dir),
'wget -S --progress=dot:giga -O- {} | tar -xz -C {}'.format(args.mesa_build_url, args.ci_project_dir),
'wget -S --progress=dot:giga -O- {} | tar -xz -C /'.format(args.job_rootfs_overlay_url),
f'echo "export CI_JOB_JWT_FILE={args.jwt_file}" >> /set-job-env-vars.sh',
'exec /init-stage2.sh',
]
test['definitions'][0]['repository']['run']['steps'] = init_lines
values['actions'] = [
{ 'deploy': deploy },
{ 'boot': boot },
{ 'test': test },
]
return yaml.dump(values, width=10000000)
def setup_lava_proxy():
config = lavacli.load_config("default")
uri, usr, tok = (config.get(key) for key in ("uri", "username", "token"))
uri_obj = urllib.parse.urlparse(uri)
uri_str = "{}://{}:{}@{}{}".format(uri_obj.scheme, usr, tok, uri_obj.netloc, uri_obj.path)
transport = lavacli.RequestsTransport(
uri_obj.scheme,
config.get("proxy"),
config.get("timeout", 120.0),
config.get("verify_ssl_cert", True),
)
proxy = xmlrpc.client.ServerProxy(
uri_str, allow_none=True, transport=transport)
print_log("Proxy for {} created.".format(config['uri']))
return proxy
def _call_proxy(fn, *args):
retries = 60
for n in range(1, retries + 1):
try:
return fn(*args)
except xmlrpc.client.ProtocolError as err:
if n == retries:
traceback.print_exc()
fatal_err("A protocol error occurred (Err {} {})".format(err.errcode, err.errmsg))
else:
time.sleep(15)
except xmlrpc.client.Fault as err:
traceback.print_exc()
fatal_err("FATAL: Fault: {} (code: {})".format(err.faultString, err.faultCode))
def get_job_results(proxy, job_id, test_suite, test_case):
# Look for infrastructure errors and retry if we see them.
results_yaml = _call_proxy(proxy.results.get_testjob_results_yaml, job_id)
results = yaml.load(results_yaml, Loader=loader(False))
for res in results:
metadata = res["metadata"]
if "result" not in metadata or metadata["result"] != "fail":
continue
if 'error_type' in metadata and metadata['error_type'] == "Infrastructure":
print_log("LAVA job {} failed with Infrastructure Error. Retry.".format(job_id))
return False
if 'case' in metadata and metadata['case'] == "validate":
print_log("LAVA job {} failed validation (possible download error). Retry.".format(job_id))
return False
results_yaml = _call_proxy(proxy.results.get_testcase_results_yaml, job_id, test_suite, test_case)
results = yaml.load(results_yaml, Loader=loader(False))
if not results:
fatal_err("LAVA: no result for test_suite '{}', test_case '{}'".format(test_suite, test_case))
print_log("LAVA: result for test_suite '{}', test_case '{}': {}".format(test_suite, test_case, results[0]['result']))
if results[0]['result'] != 'pass':
fatal_err("FAIL")
return True
def wait_until_job_is_started(proxy, job_id):
print_log(f"Waiting for job {job_id} to start.")
current_state = "Submitted"
waiting_states = ["Submitted", "Scheduling", "Scheduled"]
while current_state in waiting_states:
job_state = _call_proxy(proxy.scheduler.job_state, job_id)
current_state = job_state["job_state"]
time.sleep(WAIT_FOR_DEVICE_POLLING_TIME_SEC)
print_log(f"Job {job_id} started.")
def follow_job_execution(proxy, job_id):
line_count = 0
finished = False
last_time_logs = datetime.now()
while not finished:
(finished, data) = _call_proxy(proxy.scheduler.jobs.logs, job_id, line_count)
if logs := yaml.load(str(data), Loader=loader(False)):
# Reset the timeout
last_time_logs = datetime.now()
for line in logs:
print("{} {}".format(line["dt"], line["msg"]))
line_count += len(logs)
else:
time_limit = timedelta(seconds=DEVICE_HANGING_TIMEOUT_SEC)
if datetime.now() - last_time_logs > time_limit:
print_log("LAVA job {} doesn't advance (machine got hung?). Retry.".format(job_id))
return False
# `proxy.scheduler.jobs.logs` does not block, even when there is no
# new log to be fetched. To avoid dosing the LAVA dispatcher
# machine, let's add a sleep to save them some stamina.
time.sleep(LOG_POLLING_TIME_SEC)
return True
def show_job_data(proxy, job_id):
show = _call_proxy(proxy.scheduler.jobs.show, job_id)
for field, value in show.items():
print("{}\t: {}".format(field, value))
def validate_job(proxy, job_file):
try:
return _call_proxy(proxy.scheduler.jobs.validate, job_file, True)
except:
return False
def submit_job(proxy, job_file):
return _call_proxy(proxy.scheduler.jobs.submit, job_file)
def retriable_follow_job(proxy, yaml_file):
retry_count = NUMBER_OF_RETRIES_TIMEOUT_DETECTION
while retry_count >= 0:
job_id = submit_job(proxy, yaml_file)
print_log("LAVA job id: {}".format(job_id))
wait_until_job_is_started(proxy, job_id)
if not follow_job_execution(proxy, job_id):
print_log(f"Job {job_id} has timed out. Cancelling it.")
# Cancel the job as it is considered unreachable by Mesa CI.
proxy.scheduler.jobs.cancel(job_id)
retry_count -= 1
continue
show_job_data(proxy, job_id)
if get_job_results(proxy, job_id, "0_mesa", "mesa") == True:
break
else:
# The script attempted all the retries. The job seemed to fail.
return False
return True
def main(args):
proxy = setup_lava_proxy()
yaml_file = generate_lava_yaml(args)
if args.dump_yaml:
print(hide_sensitive_data(generate_lava_yaml(args)))
if args.validate_only:
ret = validate_job(proxy, yaml_file)
if not ret:
fatal_err("Error in LAVA job definition")
print("LAVA job definition validated successfully")
return
if not retriable_follow_job(proxy, yaml_file):
fatal_err(
"Job failed after it exceeded the number of"
f"{NUMBER_OF_RETRIES_TIMEOUT_DETECTION} retries."
)
def create_parser():
parser = argparse.ArgumentParser("LAVA job submitter")
parser.add_argument("--pipeline-info")
parser.add_argument("--base-system-url-prefix")
parser.add_argument("--mesa-build-url")
parser.add_argument("--job-rootfs-overlay-url")
parser.add_argument("--job-artifacts-base")
parser.add_argument("--job-timeout", type=int)
parser.add_argument("--first-stage-init")
parser.add_argument("--ci-project-dir")
parser.add_argument("--device-type")
parser.add_argument("--dtb", nargs='?', default="")
parser.add_argument("--kernel-image-name")
parser.add_argument("--kernel-image-type", nargs='?', default="")
parser.add_argument("--boot-method")
parser.add_argument("--lava-tags", nargs='?', default="")
parser.add_argument("--jwt-file", type=pathlib.Path)
parser.add_argument("--validate-only", action='store_true')
parser.add_argument("--dump-yaml", action='store_true')
parser.add_argument("--visibility-group")
return parser
if __name__ == "__main__":
# given that we proxy from DUT -> LAVA dispatcher -> LAVA primary -> us ->
# GitLab runner -> GitLab primary -> user, safe to say we don't need any
# more buffering
sys.stdout.reconfigure(line_buffering=True)
sys.stderr.reconfigure(line_buffering=True)
parser = create_parser()
parser.set_defaults(func=main)
args = parser.parse_args()
args.func(args)
Reference in New Issue View Git Blame Copy Permalink