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
Currently, I am avoiding using this library, as I want to be able to prioritize write operations to go directly to the leader node (if we have a healthy one). Right now it seems the only logic that happens when making API requests, is retry logic to make sure there is a healthy node, which is good, but it also means writes can go to a follower node, which is inefficient and then has to be redirected to the writer node.
Expected Behavior
Node class should encapsulate whether or not a node is the writer. Ideally this would be cached from the /debug endpoint 'state=1' value. If there is a healthy writer node available, use it for write operations to a collection.
Metadata
Typsense Version: 0.19
OS: Linux
The text was updated successfully, but these errors were encountered:
@jasonbosco can you explain why this was marked as wontfix? Just wondering what kind of a performance impact this has, evaluating using the REST API vs this library.
Description
Currently, I am avoiding using this library, as I want to be able to prioritize write operations to go directly to the leader node (if we have a healthy one). Right now it seems the only logic that happens when making API requests, is retry logic to make sure there is a healthy node, which is good, but it also means writes can go to a follower node, which is inefficient and then has to be redirected to the writer node.
Expected Behavior
Node class should encapsulate whether or not a node is the writer. Ideally this would be cached from the /debug endpoint 'state=1' value. If there is a healthy writer node available, use it for write operations to a collection.
Metadata
Typsense Version: 0.19
OS: Linux
The text was updated successfully, but these errors were encountered: