Add IPv4 prefix set integration test #251
Conversation
I suspect the reason for the "update is stale" is due to a race condition between different calls to gNMI.Set that each may provide a different unix.Now() value depending on when it is executed prior to updating the cache. I believe the correct fix is to actually to ignore it. WIP -- isn't Set synchronous? If so how could this happen? clue: is it always happening on default values?
|
What's the long term plan for this? It seems like there's a lot duplication (code and test behavior) of the local test. |
I agree there is a lot of duplication. I think an Ondatra binding or generics/interfaces may be the solution to deal with the different Device/Port/ygnmi API type differences, as otherwise the two are identical. My judgement is that local_tests will be useful as a lightweight way to test and debug BGP policy development, so once the Ondatra version of the test is stable, I'll look at an Ondatra bind for local lemmings. |
Maybe fix for "update is stale"
Skip Initial Bad packets in BGP-triggered GUE test
|
@DanG100 let me know if you have more comments on this one. This PR contains the changes that reduces test flakiness due to chaining the PRs. |
2 participants
Previous PR: #253