feat(idempotency): handle lambda timeout scenarios for INPROGRESS records

[rubenfonseca]

Issue number: #1038

Summary

Expire in-progress invocations caused by Lambda time outs.

Changes

Please provide a summary of what's being changed

During an invocation, we try to figure out how much time the Lambda invocation has before it expires, and save it on the idempotency record together will all the other fields.

When a new invocation with the same key arrives, we check against this timestamp to see if there's any INPROGRESS invocation past the timeout timestamp. If there is, we allow the invocation to be retried.

We also moved the squence diagrams of the idempotency module in documentation to mermaid.js.

Caveats

• Only works out of the box for the @idempotent decorator, because right now it's the only way to access the Lambda context to get the remaining time left. I've added a register_lambda_context to the IdempotencyConfig class as a workaround

User experience

Please share what the user experience looks like before and after this change

Before: if a Lambda invocation times out, the invocation will get stuck until the idempotency record expires (default = 1 hour).

After: if a Lambda invocation times out, the invocation will be tried again on the next retry.

Checklist

If your change doesn't seem to apply, please leave them unchecked.

• Meet tenets criteria
• I have performed a self-review of this change
• Changes have been tested
• Changes are documented
• PR title follows conventional commit semantics

Is this a breaking change?

RFC issue number:

Checklist:

• Migration process documented
• Implement warnings (if it can live side by side)

Acknowledgment

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

Disclaimer: We value your time and bandwidth. As such, any pull requests created on non-triaged issues might not be successful.

---

View rendered docs/utilities/idempotency.md

[codecov-commenter]

Codecov Report

Merging #1387 (66b62a6) into develop (7414df7) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff            @@
##           develop    #1387   +/-   ##
========================================
  Coverage    99.88%   99.89%           
========================================
  Files          119      119           
  Lines         5429     5456   +27     
  Branches       620      624    +4     
========================================
+ Hits          5423     5450   +27     
  Misses           2        2           
  Partials         4        4           

Impacted Files	Coverage Δ
...ws_lambda_powertools/utilities/idempotency/base.py	100.00% <100.00%> (ø)
..._lambda_powertools/utilities/idempotency/config.py	100.00% <100.00%> (ø)
...da_powertools/utilities/idempotency/idempotency.py	100.00% <100.00%> (ø)
...wertools/utilities/idempotency/persistence/base.py	99.39% <100.00%> (+0.02%)	⬆️
...ools/utilities/idempotency/persistence/dynamodb.py	98.80% <100.00%> (+0.10%)	⬆️
aws_lambda_powertools/tracing/base.py	100.00% <0.00%> (ø)
aws_lambda_powertools/event_handler/appsync.py	100.00% <0.00%> (ø)
aws_lambda_powertools/event_handler/__init__.py	100.00% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c463232...66b62a6. Read the comment docs.

[heitorlessa]

early comments before I resume reading it after lunch. At a quick glance, the pieces that would benefit the most now are:

• 1/ Comments on the formula you're using since I saw Lambda context giving us milliseconds, then epoch (sensible for comparison), then type coercion to string (?) - sounds like we could reuse the milliseconds instead of wall clock time. Also, datetime.datetime.now() uses timezone offset too, instead of a safe non-backward timestamp like time.monotonic
• 2/ I think the Update Expression would benefit to having the new one as a separate variable, then use f-string to combine them and make it more readable (which conditional is for which use case)

[heitorlessa]

I've pushed an updated doc with the following changes:

• Reordered the wording so we give the positive news first (what exactly we do we do?)
• Give comfort to those using @idempotent (When do they need to care?)
• Used banners to emphasize when they need to take action (what should I do?)
• Moved sequence diagram for latter so they're informed first, then they can dig deeper into the mechanics

[heitorlessa]

Shortening the wording on sequence diagram to fit in the final SVG

New (pushing shortly)