CloudFormation Events Timeline view in a Gantt Chart

AWS released a CloudFormation Timeline to the AWS Console. Presenting the Stack Events like this helps us comprehend the reasons why a Stack takes time to update, and visualize causes of failure.

GitHub Actions runs my deployments, code that gets merged into main runs in GitHub Actions. That's where I'm looking at when I click merge Pull Request. When there is an error debugging is

Going from GitHub Acions to the AWS Console.

Give me the Stack Name

from argparse import ArgumentParser

parser = ArgumentParser()
parser.add_argument("stack", help="Name of CloudFormation Stack")

# tbh i could find this, but i already have this vslue where I use this
parser.add_argument("start", help="Time of First in Event ISO Formatted Time")

args = parser.parse_args()

init_time = datetime.fromisoformat(args.start)
stack_name = args.stack

Get the Stack Events

from boto3 import Session
from datetime import datetime

session = Session()
cloudformation = session.client("cloudformation")
paginator = cloudformation.get_paginator("describe_stack_events")


def get_deployment_events(stack_name: str, start_time: datetime):
    for page in paginator.paginate(StackName=stack_name):
        for stack_event in page["StackEvents"]:
            if stack_event["Timestamp"] < start_time:
                return
            yield stack_event

Group Events By Resource

from collections import defaultdict

events_by_resource = defaultdict(list)
for event in get_deployment_events(stack_name, init_time):
    events_by_resource[event["LogicalResourceId"]].append(event)

Build a Gantt Chart

MERMAID_GANTT = f"""
gantt
    title Stack {stack_name} Deployment
    dateFormat x
    axisFormat %S.%L
"""

for section_name, events in events_by_resource.items():
    section = f"\n    section {section_name} [{events[0]["ResourceType"]}]\n"
    MERMAID_GANTT += section

    for index, event in enumerate(events):
        status = event["ResourceStatus"]
        earlier_events = events[index+1:]

        if status.rpartition("_")[-1] not in {"FAILED", "COMPLETE"}:
            continue

        init_status = "_".join((status.rpartition("_")[0], "IN_PROGRESS"))
        end_time = int(event["Timestamp"].timestamp() * 1000)

        # try to find init time
        init_time = end_time  # use end_time as default if can't find start
        # there may be more than 1 event with the init_status
        # we naively at the moment select the earliest
        candidates = [candidate for candidate in earlier_events if candidate["ResourceStatus"] == init_status]
        if candidates:
            earliest_candidate = candidates[-1]
            init_time = int(earlist_candidate["Timestamp"].timestamp() * 1000)

        is_failure = status.endswith("FAILED")
        label = "crit" if is_failure else "done"
        MERMAID_GANTT += f"    {status} :{label}, {init_time}, {end_time}\n"


def format_github(mermaid_diagram_data):
    return f"```mermaid\n{mermaid_diagram_data}\n```"


print(format_github(MERMAID_GANTT))

And the output

gantt
    title Stack Example Deployment
    dateFormat x
    axisFormat %S.%L

    section Example [AWS::CloudFormation::Stack]
    CREATE_COMPLETE :done, 1732724310778, 1732724331438

    section RoleB [AWS::IAM::Role]
    CREATE_COMPLETE :done, 1732724312879, 1732724330031

    section RoleA [AWS::IAM::Role]
    CREATE_COMPLETE :done, 1732724312847, 1732724329912

    section PolicyA [AWS::IAM::ManagedPolicy]
    CREATE_COMPLETE :done, 1732724312866, 1732724329101