fix(AWS SQS): ensure sqs event depends on provisioned alias if needed

[pgrzesik]

Ensure that EventSourceMapping depends on provisioned alias if it exists.

One question: do you think it makes sense to add an integration test with provisionedConcurrency as well? I reproduced/tested the fix based on adjusted existing SQS integration test which made me wonder about adding it permanently.

Closes: #8249

[codecov-commenter]

Codecov Report

Merging #8298 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #8298   +/-   ##
=======================================
  Coverage   88.16%   88.17%           
=======================================
  Files         250      250           
  Lines        9391     9397    +6     
=======================================
+ Hits         8280     8286    +6     
  Misses       1111     1111           

Impacted Files	Coverage Δ
...ib/plugins/aws/package/compile/events/sqs/index.js	100.00% <100.00%> (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 e990c09...bc3c66a. Read the comment docs.

[medikoo]

Thank you @pgrzesik !

One question: do you think it makes sense to add an integration test with provisionedConcurrency as well?

I wanted to do that, at some point, and even had some test ready, still it was very slow. It seems that provisioned concurrency implies a slow deploy and remove time.

What test evaluation times did you observe (?)

[pgrzesik]

Hey @medikoo 👋

I ran the tests about ~10 times for both provisioned and non-provisioned and I observed:

• ~3-4 minutes for non-provisioned
• ~7 minutes for provisioned with provisionedConcurrency: 1

So while it's more time consuming, I feel like it's still within reasonable limits. What do you think?

[medikoo]

@pgrzesik thanks for sharing, I think when I tried this (shortly after provisioned conncurrency was introduced) it was around 15 minutes, so it probably got improved.

In light of that, yes I totally agree it's worth to have some deployment that tests provisionedConcurrency setup. It'll probably be optimal to have one dedicated stack (test), for testing all provisioned concurrency cases (but we can start just with this one). What do you think?

[pgrzesik]

Hey there @medikoo - sorry for not responding sooner - I've decided to move the discussion to an issue, hope that's okay - #8321