Alert: XDM activity

[teor2345]

Alert on:

• an XDM above a threshold - part of the transfer value check
• total transfer volume over a few days
• large numbers of XDMs

This might be able to be implemented using:

• storage or RPC/Runtime API calls

[teor2345]

@vedhavyas do you think we'll need any extra runtime calls for these XDM alerts?

(Let's assume we can get the arguments to and from the node using the correct types, that seems to be working well.)

[vedhavyas]

I'm not sure about these alerts specifically how the checks are done.

• an XDM above a threshold

What XDM above threshold here means ?

• total transfer volume over a few days

This might be seems to require some form of storage layer either in-memory or persistant. Is there some kind of action we would like to take against these or these just notifications for us I assume

• large numbers of XDMs

You can use the existing channel storages to see if there are large number of XDMs initiated in a given block or is there window we want to use here ?

For me here are some of the actionable alerts

• An outboxed XDM not executed as Inbox on dst_chain post the wait_period + some_threshold_blocks
• An Inbox response XDM not executed on dst_chain post the wait_period + some_threshold_blocks
• Failed XDM Outbox Extrinsics count in a block
• Any failed inbox response XDM from dst_chain

Some good to know notifications

• Huge number of XDMs in a single block. Need to define what is Huge number of XDM here means
• Big Transfer from src_chain to dst_chain. Need to define what is Big transfer here means

For all of the alerts, we would need to track their XDM message id and dst_chain: (src_chain, dst_chain, message_type, channel_id, nonce)

do you think we'll need any extra runtime calls for these XDM alerts?

I think all of these can be done using a combination of storage and runtime-calls I believe on the first pass. Maybe there are some alerts that may require new runtime-calls but I dont see such requirement as of now

[teor2345]

• total transfer volume over a few days

This might be seems to require some form of storage layer either in-memory or persistant. Is there some kind of action we would like to take against these or these just notifications for us I assume

Just notifications. I think we can mark this as lower priority for now. We could check recent extrinsics or events on startup, to avoid storing the amounts ourselves.

• large numbers of XDMs

You can use the existing channel storages to see if there are large number of XDMs initiated in a given block or is there window we want to use here ?

I think it's easier just to focus on checking a single block for now. We could use storage, but it might be easier to use relay_message extrinsics or Inbox/Outbox events.

I think all of these can be done using a combination of storage and runtime-calls I believe on the first pass. Maybe there are some alerts that may require new runtime-calls but I dont see such requirement as of now

Happy to use storage or runtime calls if needed.