I have a serverless application, which consists of numerous lambda functions. Each lambda function has a tag “STAGE”, which is set either to “production” or “staging”. What I need now is to retrieve the data on overall invocations for each production lambda function for the last calendar month (30 days or 4 weeks), with a period of 1 day. I manage to achieve that in the AWS Console (Cloudwatch > Explorer), like shown on the screenshot below:

But the thing is that I need to retrieve this data programmatically to pass it down an ETL pipeline of sorts, that will generate a report with a specific format. I’ve came up with a following script to retrieve this data:

import datetime
import boto3

client = boto3.client("cloudwatch", region_name="eu-west-1")
from_date = datetime.date.today()
month_ago = from_date - datetime.timedelta(days=30)

response = client.get_metric_data(
    MetricDataQueries=[{
        "Id": "test",
        "MetricStat": {
            "Metric": {
                "Namespace": "AWS/Lambda",
                "MetricName": "Invocations",
                "Dimensions": [{
                    "Name": "STAGE",
                    "Value": "production"
                 }]
             },
             "Period": 86400,
             "Stat": "Sum"
          },
          "ReturnData": True,
   }],
   StartTime=month_ago.strftime("%Y-%m-%dT00:00:00Z"),
   EndTime=from_date.strftime("%Y-%m-%dT00:00:00Z")
)

But all it returns me with these params is an empty dataset: { 'MetricDataResults': [{'Id': 'test', 'Label': 'Invocations', 'Timestamps': [], 'Values': [], 'StatusCode': 'Complete'}] ...}

And trying to run with the following Expression "SELECT SUM(Invocations) FROM SCHEMA("AWS/Lambda") WHERE "STAGE" = 'production' GROUP BY FunctionName" instead of a MetricStat dict raises a MaxQueryTimeRangeExceed exception, since we’re allowed to SELECT the data only on the last 3 hours.

How would I build the query correctly to retrieve the same data with the same filters as I do in the AWS Console in this case? Thanks in advance!