You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Oct 17, 2022. It is now read-only.
On the current implementation when a worker forms a new batch it broadcasts it to the worker nodes of the same id of other validators. That assumption is propagated across the codebase - meaning that irrespective of which primary node we query, we assume that a specific batch will be found to the worker of same id (which is basically dictated by the propagated header)
The above assumption is making scaling quite restrictive and eventually we want it removed. On our network our primary node should keep the mapping it self to which local worker a batch has been stored and basically decouple the storying/query semantics from the ones that the author produces (worker id). It has to be noted that a batch is identified by two parameters: the batch digest and the worker_id (author). We still want to be able to identify batches based on both those two properties, but for our primary node we should be able to translate that to which local worker the batch has been stored.
The text was updated successfully, but these errors were encountered:
On the current implementation when a worker forms a new batch it broadcasts it to the worker nodes of the same id of other validators. That assumption is propagated across the codebase - meaning that irrespective of which primary node we query, we assume that a specific batch will be found to the worker of same id (which is basically dictated by the propagated header)
The above assumption is making scaling quite restrictive and eventually we want it removed. On our network our primary node should keep the mapping it self to which local worker a batch has been stored and basically decouple the storying/query semantics from the ones that the author produces (worker id). It has to be noted that a batch is identified by two parameters: the batch digest and the worker_id (author). We still want to be able to identify batches based on both those two properties, but for our primary node we should be able to translate that to which local worker the batch has been stored.
The text was updated successfully, but these errors were encountered: