Investigate the following:

[JoeCap08055]

Description

Investigate the following from load testing the SIWF sign-up flow in account-service:

• Why we get nonce conflicts when queue concurrency is increased
• Why we got an apparent nonce hole (that was never filled) when submitting a large number of transactions (even with queue concurrency=1)

[aramikm]

I think caching the nonce might have a side effect in causing the nonce hole since we have not experienced that issue before

[JoeCap08055]

I think caching the nonce might have a side effect in causing the nonce hole since we have not experienced that issue before

Quite possibly; though I would think it would be the opposite. Caching the nonce runs the risk of it being stale == behind, which would more likely cause nonce conflicts, not holes.

I'm working on a better solution for detecting holes...