I am trying to conceptually understand the backend architecture of a Sentinel and the details seem to be a little light. From what I understand, the only way to invoke a Autotask/Relayer script via a smart contract is through a Sentinel (is this correct?). Is the Sentinel just a scanner of sorts (ie. event listener) that scans the chosen blockchain every block?
Is AWS the single common point of failure for all (Autotask/Relayer/Sentinel)? Are there redundancies in place (ie. different AWS servers in different continents?)
If I am heading in the wrong direction here, is there a better method to invoke an autotask via a smart contract (ie. requesting data from an external api via a smart contract). I know that external adapters in Chainlink are very similar, but I need a permissioned style api call.
Hey @Guceri, excellent questions!! Thanks for asking
That's correct. You can invoke an Autotask or Relayer manually via a webhook or on a scheduled basis, but if you want to trigger them from a smart contract event, then hooking them to a Sentinel is the way to go.
Exactly! It processes every tx on every block, and for the blockchains where we have trace support (eg Ethereum Mainnet), it also analyzes internal tx traces.
At the moment, it is. We plan on going multi-region during 2022, as well as facilitating custom deployments on request. And we have a fully decentralized project in the works that may help here, so stay tuned!
We've seen some projects set up an Autotask on scheduled basis, where the autotask wakes up, checks a set of conditions in a smart contract, and based on them decides to act or not. This is not as efficient as a sentinel-based autotask, but it may work for situations where you don't have a specific function call or event to hook onto.
@spalladino - follow up to the original question. If AWS is down in the region where your sentinels are running, is there a guarantee that the event will eventually fire when they come back online? Is there an SQS queue in place or something? It seems that we would need some guarantee that the event will eventually fire to build any critical operations that utilize this service.
