Using @metric_scope with asyncio #101
Description
I have a Lambda function that gets called and it executes a lot of HTTP requests. I'm trying to speed those up through aiohttp. Ultimately I want 1 log entry per http request and 1 log for the actual lambda invocation. I'm running into problems with the metric logger decorating an async method.
Traceback (most recent call last):\n File \"/var/task/index.py\", line 174, in process_requests\n code = await send_request(request, context, session)\n File \"/var/task/aws_embedded_metrics/metric_scope/__init__.py\", line 34, in wrapper\n await logger.flush()\n File \"/var/task/aws_embedded_metrics/logger/metrics_logger.py\", line 47, in flush\n sink.accept(self.context)\n File \"/var/task/aws_embedded_metrics/sinks/stdout_sink.py\", line 25, in accept\n for serialized_content in self.serializer.serialize(context):\n File \"/var/task/aws_embedded_metrics/serializers/log_serializer.py\", line 106, in serialize\n event_batches.append(json.dumps(current_body))\n File \"/var/lang/lib/python3.9/json/__init__.py\", line 231, in dumps\n return _default_encoder.encode(obj)\n File \"/var/lang/lib/python3.9/json/encoder.py\", line 199, in encode\n chunks = self.iterencode(o, _one_shot=True)\n File \"/var/lang/lib/python3.9/json/encoder.py\", line 257, in iterencode\n return _iterencode(o, 0)\n File \"/var/lang/lib/python3.9/json/encoder.py\", line 179, in default\n raise TypeError(f'Object of type {o.__class__.__name__} '\nTypeError: Object of type method is not JSON serializable\n"
This is a basic version of my handler:
# Get event loop
loop = asyncio.get_event_loop()
@metric_scope
def handler(event, context, metrics):
start = time.perf_counter()
t = loop.run_until_complete(process_requests(event, context, metrics))
end = time.perf_counter()
metrics.put_metric("TotalProcessingLatency", (end - start) * 1000, "Milliseconds")
return tThe process_requests method:
async def process_requests(event, context, metrics):
"""
Sets up the aiohttp client and iterates all of the requests
through the session
"""
tracebacks = []
errors = []
async with aiohttp.ClientSession() as session:
for request in event["requests"]:
for x in range(0, request["iterations"]):
try:
code = await send_request(request, context, session)
except Exception as e:
errors.append(str(e))
tracebacks.append(traceback.format_exc())
metrics.set_property("Errors", errors)
metrics.set_property("Tracebacks", tracebacks)
return NoneAnd finally, the send_request method:
@metric_scope
async def send_request(request, context, session, metrics):
start = time.time() * 1000
metrics.set_property("RequestStartTime", round(start))
async with session.request(method = request["method"],
headers = h,
url = request["url"],
data = str(request["data"]),
ssl = verify,
timeout = request["timeout"]) as response:
response_end = time.time() * 1000
metrics.set_property("ResponseReceivedTime", round(response_end))
metrics.put_metric("RequestLatency", response_end - start, "Milliseconds")
return response.status_codeThis works when I process the HTTP requests synchronously, I get a log for each time send_request is called and once for the overall handler execution. Is there a way I can use the metrics logger and get the same outcome with aiohttp/asyncio?