[Crypto] improve flaky random test #2509
Conversation
Codecov Report
@@ Coverage Diff @@
## master #2509 +/- ##
=======================================
Coverage 56.93% 56.93%
=======================================
Files 656 656
Lines 60328 60328
=======================================
Hits 34349 34349
Misses 23033 23033
Partials 2946 2946
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
|
If I have the correct test (TestUint), this test has been consistently passing in the Flaky Test Monitor:
Where have you seen the flaky behavior? |
|
@gomisha this test started failed a few times on feature branch for Active Pacemaker. Here one if the evidences: https://github.com/onflow/flow-go/runs/6622170032?check_suite_focus=true |
|
The test is probabilistic and there is always a low chance of failure, failures should happen very rarely. I'm also surprised it happens "a few times" on the feature branch. |
| crand.Read(seed) | ||
| rand.Read(seed) |
There was a problem hiding this comment.
why replacing crypto/rand with math/rand?
There was a problem hiding this comment.
math/rand produces deterministic randoms depending on the initializing seed, while crypto/rand doesn't allow that. This is useful when a test fails, and by logging the seed we can reproduce the fail.
Note that math/rand shouldn't be used outside of tests when secure entropy is needed.
|
bors merge |
A rare flakiness has been seen in a statistical tests under crypto/random
cc @durkmurder