Many tests seem to fail randomly in our test-suite. It seems that due to large number of CPI calls, we are running into compute units during some runs of the contract call and not on another due to several reasons such as
- Randomness when Solana finds the discriminator of our created keypair
- Randomness assosciated with low level APIs depending on slot.
etc.
Overall, we should try to isolate which calls then to fail and add a pre-instruction to them to add compute units, the gas will be refunded if they are not needed, else they will be used and the call will not fail, reducing randomness in our tests.
This issue was created at https://www.defi-os.com , please login/signup on it to get cash rewards for solving it.
Many tests seem to fail randomly in our test-suite. It seems that due to large number of CPI calls, we are running into compute units during some runs of the contract call and not on another due to several reasons such as
etc.
Overall, we should try to isolate which calls then to fail and add a pre-instruction to them to add compute units, the gas will be refunded if they are not needed, else they will be used and the call will not fail, reducing randomness in our tests.
This issue was created at https://www.defi-os.com , please login/signup on it to get cash rewards for solving it.