Delete the wasm cache if pod is crashlooping

[DavidNix]

This gets into a larger self-healing feature of crashloop mitigation.

See: CosmWasm/wasmvm#426

We could delete the cache on every restart, but I'm unsure how it will affect performance. Per the above, there will be a "slowdown".

[webmaster128]

We could delete the cache on every restart, but I'm unsure how it will affect performance. Per the above, there will be a "slowdown".

I'd definitely recommend against that. You'd have to re-compile Wasm -> native for each contract after every start.

If you do it once per chain upgrade, that's safe enough. But at the end of the day that's a CosmWasm bug that will be fixed in 1.3 and beyond.

[DavidNix]

We could delete the cache on every restart, but I'm unsure how it will affect performance. Per the above, there will be a "slowdown".

I'd definitely recommend against that. You'd have to re-compile Wasm -> native for each contract after every start.

If you do it once per chain upgrade, that's safe enough. But at the end of the day that's a CosmWasm bug that will be fixed in 1.3 and beyond.

Thanks for the feedback @webmaster128. Can you point me to documentation or discussion that shows it will be fixed in 1.3 and beyond?

[webmaster128]

See CosmWasm/wasmvm#426 (comment)

[DavidNix]

Great! If we decide to move forward with this feature, it will be opt-in and only if the pod is crashlooping. We won't automatically blow away the cache on every restart.

[DavidNix]

Rare edge case. Automation seems overkill, so closing.