fixed definition order broken by auto-refac; reduced prometheus code to only forwarding to analysis endpoint

This commit is contained in:
Matthias Bisping 2022-03-16 11:16:39 +01:00
parent 9f3c884c75
commit 894a6b5d4c
4 changed files with 43 additions and 61 deletions

View File

@ -2,9 +2,8 @@ import logging
import requests
from flask import Flask, jsonify
from retry import retry
from waitress import serve
from prometheus_client import generate_latest, CollectorRegistry
from pyinfra.config import CONFIG
@ -42,11 +41,6 @@ def set_up_probing_webserver():
return resp
@app.route("/prometheus", methods=["GET"])
def prometheus():
analysis_metrics = get_metrics_from_analysis_endpoint()
combined_metrics = f"{analysis_metrics}{generate_latest(PROM_REGISTER).decode()}"
return combined_metrics
def get_metrics_from_analysis_endpoint():
try:
metric = requests.get(f"{CONFIG.rabbitmq.callback.analysis_endpoint}/prometheus")
@ -54,9 +48,8 @@ def set_up_probing_webserver():
return metric.text
except requests.exceptions.ConnectionError as err:
logging.warning(f"Got no metrics from analysis prometheus endpoint: {err}")
return "# WARNING No Analysis metrics\n"
resp = jsonify("no analysis metrics received")
resp.status_code = 504
return resp
return app
PROM_REGISTER = CollectorRegistry()

View File

@ -26,6 +26,16 @@ def parse_args():
return args
def combine_dossier_id_and_file_id_and_extension(dossier_id, file_id, extension):
return f"{dossier_id}/{file_id}{extension}"
def upload_compressed_response(storage, bucket_name, dossier_id, file_id, result) -> None:
data = gzip.compress(result.encode())
path_gz = combine_dossier_id_and_file_id_and_extension(dossier_id, file_id, CONFIG.service.response.extension)
storage.put_object(bucket_name, path_gz, data)
def add_file_compressed(storage, bucket_name, dossier_id, path) -> None:
path_gz = combine_dossier_id_and_file_id_and_extension(dossier_id, Path(path).stem, ".ORIGIN.pdf.gz")
@ -57,13 +67,3 @@ if __name__ == "__main__":
elif args.command == "purge":
storage.clear_bucket(bucket_name)
def combine_dossier_id_and_file_id_and_extension(dossier_id, file_id, extension):
return f"{dossier_id}/{file_id}{extension}"
def upload_compressed_response(storage, bucket_name, dossier_id, file_id, result) -> None:
data = gzip.compress(result.encode())
path_gz = combine_dossier_id_and_file_id_and_extension(dossier_id, file_id, CONFIG.service.response.extension)
storage.put_object(bucket_name, path_gz, data)

View File

@ -6,6 +6,34 @@ from pyinfra.config import CONFIG
from pyinfra.storage.storages import get_s3_storage
def read_connection_params():
credentials = pika.PlainCredentials(CONFIG.rabbitmq.user, CONFIG.rabbitmq.password)
parameters = pika.ConnectionParameters(
host=CONFIG.rabbitmq.host,
port=CONFIG.rabbitmq.port,
heartbeat=CONFIG.rabbitmq.heartbeat,
credentials=credentials,
)
return parameters
def make_channel(connection) -> pika.adapters.blocking_connection.BlockingChannel:
channel = connection.channel()
channel.basic_qos(prefetch_count=1)
return channel
def declare_queue(channel, queue: str):
args = {"x-dead-letter-exchange": "", "x-dead-letter-routing-key": CONFIG.rabbitmq.queues.dead_letter}
return channel.queue_declare(queue=queue, auto_delete=False, arguments=args)
def make_connection() -> pika.BlockingConnection:
parameters = read_connection_params()
connection = pika.BlockingConnection(parameters)
return connection
def build_message_bodies():
storage = get_s3_storage()
for bucket_name, pdf_name in storage.get_all_object_names(CONFIG.storage.bucket):
@ -21,23 +49,6 @@ def build_message_bodies():
).encode()
def make_channel(connection) -> pika.adapters.blocking_connection.BlockingChannel:
channel = connection.channel()
channel.basic_qos(prefetch_count=CONFIG.rabbitmq.prefetch_count)
return channel
def declare_queue(channel, queue: str):
args = {"x-dead-letter-exchange": "", "x-dead-letter-routing-key": CONFIG.rabbitmq.queues.dead_letter}
return channel.queue_declare(queue=queue, auto_delete=False, arguments=args)
def make_connection() -> pika.BlockingConnection:
parameters = read_connection_params()
connection = pika.BlockingConnection(parameters)
return connection
if __name__ == "__main__":
connection = make_connection()
@ -52,14 +63,3 @@ if __name__ == "__main__":
for method_frame, _, body in channel.consume(queue=CONFIG.rabbitmq.queues.output):
print(f"Received {json.loads(body)}")
channel.basic_ack(method_frame.delivery_tag)
def read_connection_params():
credentials = pika.PlainCredentials(CONFIG.rabbitmq.user, CONFIG.rabbitmq.password)
parameters = pika.ConnectionParameters(
host=CONFIG.rabbitmq.host,
port=CONFIG.rabbitmq.port,
heartbeat=CONFIG.rabbitmq.heartbeat,
credentials=credentials,
)
return parameters

View File

@ -1,14 +1,12 @@
import logging
import tracemalloc
from multiprocessing import Process
import requests
from prometheus_client import Gauge
from retry import retry
from pyinfra.config import CONFIG, make_art
from pyinfra.exceptions import AnalysisFailure, ConsumerError
from pyinfra.flask import run_probing_webserver, set_up_probing_webserver, PROM_REGISTER
from pyinfra.flask import run_probing_webserver, set_up_probing_webserver
from pyinfra.queue.consumer import Consumer
from pyinfra.queue.queue_manager.pika_queue_manager import PikaQueueManager
from pyinfra.storage.storages import get_storage
@ -39,12 +37,6 @@ def make_callback(analysis_endpoint):
def main():
if CONFIG.service.monitoring_enabled:
tracemalloc.start()
memory_peak_metric = Gauge(
"pyinfra_memory_peak_per_call", "Pyinfra memory peak when processing a message in Mb"
)
PROM_REGISTER.register(memory_peak_metric)
webserver = Process(target=run_probing_webserver, args=(set_up_probing_webserver(),))
logging.info(make_art())
@ -63,9 +55,6 @@ def main():
try:
consumer = Consumer(visitor, queue_manager)
consumer.consume_and_publish()
if CONFIG.service.monitoring_enabled:
_, peak = tracemalloc.get_traced_memory()
memory_peak_metric.set(peak / 10**6)
except Exception as err:
raise ConsumerError from err