5d611d5fae
Merge in RR/image-prediction from RED-4329-add-prometheus to master
Squashed commit of the following:
commit 7fcf256c5277a3cfafcaf76c3116e3643ad01fa4
Merge: 8381621 c14d00c
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Jun 21 15:41:14 2022 +0200
Merge branch 'master' into RED-4329-add-prometheus
commit 8381621ae08b1a91563c9c655020ec55bb58ecc5
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Jun 21 15:24:50 2022 +0200
add prometheus endpoint
commit 26f07088b0a711b6f9db0974f5dfc8aa8ad4e1dc
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Jun 21 15:14:34 2022 +0200
refactor
commit c563aa505018f8a14931a16a9061d361b5d4c383
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Jun 21 15:10:19 2022 +0200
test bamboo build
commit 2b8446e703617c6897b6149846f2548ec292a9a1
Author: Julius Unverfehrt <julius.unverfehrt@iqser.com>
Date: Tue Jun 21 14:40:44 2022 +0200
RED-4329 add prometheus endpoint with summary metric
85 lines
2.4 KiB
Python
85 lines
2.4 KiB
Python
import multiprocessing
|
|
import traceback
|
|
from typing import Callable
|
|
|
|
from flask import Flask, request, jsonify
|
|
from prometheus_client import generate_latest, CollectorRegistry, Summary
|
|
|
|
from image_prediction.utils import get_logger
|
|
|
|
logger = get_logger()
|
|
|
|
|
|
def run_in_process(func):
|
|
p = multiprocessing.Process(target=func)
|
|
p.start()
|
|
p.join()
|
|
|
|
|
|
def wrap_in_process(func_to_wrap):
|
|
def build_function_and_run_in_process(*args, **kwargs):
|
|
def func():
|
|
try:
|
|
result = func_to_wrap(*args, **kwargs)
|
|
return_dict["result"] = result
|
|
except:
|
|
logger.error(traceback.format_exc())
|
|
|
|
manager = multiprocessing.Manager()
|
|
return_dict = manager.dict()
|
|
run_in_process(func)
|
|
return return_dict.get("result", None)
|
|
|
|
return build_function_and_run_in_process
|
|
|
|
|
|
def make_prediction_server(predict_fn: Callable):
|
|
app = Flask(__name__)
|
|
registry = CollectorRegistry(auto_describe=True)
|
|
metric = Summary(
|
|
f"redactmanager_imageClassification_seconds", f"Time spent on image-service classification.", registry=registry
|
|
)
|
|
|
|
@app.route("/ready", methods=["GET"])
|
|
def ready():
|
|
resp = jsonify("OK")
|
|
resp.status_code = 200
|
|
return resp
|
|
|
|
@app.route("/health", methods=["GET"])
|
|
def healthy():
|
|
resp = jsonify("OK")
|
|
resp.status_code = 200
|
|
return resp
|
|
|
|
def __failure():
|
|
response = jsonify("Analysis failed")
|
|
response.status_code = 500
|
|
return response
|
|
|
|
@app.route("/predict", methods=["POST"])
|
|
@app.route("/", methods=["POST"])
|
|
@metric.time()
|
|
def predict():
|
|
|
|
# Tensorflow does not free RAM. Workaround: Run prediction function (which instantiates a model) in sub-process.
|
|
# See: https://stackoverflow.com/questions/39758094/clearing-tensorflow-gpu-memory-after-model-execution
|
|
predict_fn_wrapped = wrap_in_process(predict_fn)
|
|
|
|
logger.info("Analysing...")
|
|
predictions = predict_fn_wrapped(request.data)
|
|
|
|
if predictions is not None:
|
|
response = jsonify(predictions)
|
|
logger.info("Analysis completed.")
|
|
return response
|
|
else:
|
|
logger.error("Analysis failed.")
|
|
return __failure()
|
|
|
|
@app.route("/prometheus", methods=["GET"])
|
|
def prometheus():
|
|
return generate_latest(registry=registry)
|
|
|
|
return app
|