mesh-fetch-message-chunk gets killed when large files are received

[bob-walker-nhs]

Sending a text file of 60,442,799 bytes causes the mesh-fetch-message-chunk Lambda to get killed by the AWS execution environment. e.g.:

2024-03-25T16:34:09.475+00:00
RequestId: e400f172-58e6-4a38-9b67-1e057be6fbe8 Error: Runtime exited with error: signal: killed
Runtime.ExitError

This then becomes unresolvable, because the message never gets acknowledged, and the mesh-get-messages Lambda continues to identify the offending message as available. The only way I could resolve it was to log into MESHUI and download the message so that it would get acked. This is probably not as easy in Production. Whilst in this state, our system is unusable.

Tracking here to ask the questions:

• Can/should this be handled in the client?
• Is the answer just to allocate more memory to this Lambda?
• is there an easier way to resolve it if it occurs in Production than acking the message in MESHUI?

[bob-walker-nhs]

I don't know the threshold of where this happens. A file of 16,731,558 bytes worked fine, so it's somewhere between those numbers, appreciate that doesn't narrow it down enough.

[mikewongblinx]

we face same problem by resolving increase memory of lambda now. Any better suggestion? Beside the terraform is not allow us to control lambda memory size. Is it by purpose or just simply missing variable ?