Warn instead of trace when errors happen in the spin_redis_engine

[karthik2804]

Changes the tracing level to warn as otherwise it leads to cases where the module errored and it looks like Spin hanged. I ran into this while using an invalid component. I could not understand its cause until @dicej recommended switching from debug to trace.

[dicej]

LGTM.

This does raise the question of why we're silently dropping errors here: https://github.com/fermyon/spin/blob/main/crates/redis/src/lib.rs#L97 -- there's a non-trivial amount of code dedicated to tracking and collecting errors in that file, and yet we end up just swallowing them completely.

[karthik2804]

@dicej what would be the recommended way of handling it? add some tracing there?

[dicej]

I would suggest doing a tracing::warn in the RedisTrigger::run loop instead of dropping the error (which would make the change you already made unnecessary). Either that, or stick with your current change and remove the Result<()> return types from RedisTrigger::handle and RedisExecutor::execute since they're just ignored anyway.

[itowlson]

Moving alerting up to the top level would also flag errors that happen before the Wasm module gets called - looks like there are places during load/instantiate where that could happen?

[karthik2804]

@dicej @itowlson - I have updated the PR.