For https://github.com/momentohq/dev-eco-issue-tracker/issues/536

I have been observing failed CIs on this repo since a while and the most common culprits in my observation have been:

• 9/10 times it was the test in this PR that failed. The test essentially checks that the expiration duration of a token generated at time x is within y-1 and y+1 where y is the expiration time returned by the server after the API call. The window here is small even though it does succeed locally in a loop of 1000 runs. I believe the reason it fails on CIs so frequently is either:
  • Event loop contention
  • Server side delay that might happen due to several things.
  • In one of my failed CIs, the drift was as high as 23 seconds

generate auth token using session token credentials › should succeed for generating an api token that expires

    expect(received).toBeWithin(expected)

    Expected number to be within start (inclusive) and end (exclusive):
      start: 1698452568  end: 1698452571
    Received:
      1698452594

      79 |       expect(expireResponseSuccess.is_success);
      80 |       expect(expireResponseSuccess.expiresAt.doesExpire());
    > 81 |       expect(expireResponseSuccess.expiresAt.epoch()).toBeWithin(
         |                                                       ^
      82 |         expiresIn - 1,
      83 |         expiresIn + 2
      84 |       );

Instead of testing expirationTime at a second granularity with 1 second error margin, I have changed it to test at minute granularity with 1 minute error margin. This gives some breathing room for the CIs to succeed.

• The remaining errors i have seen are due to client side timeouts. I have increased the timeout to 60 seconds from 5 seconds for similar reasons of event loop contention (just a theory).