Currently z_getoperationstatus is a potentially expensive operation. When a ztx is in the process of creation, z_getoperationstatus will block, waiting until the ztx is submitted to the mempool to return data.
The SD GUI constantly calls z_getoperationstatus and throws away any errors, which means it's constantly bombarding the full node with these calls if hushd is spending minutes on a complex tx.
This whole process is generally wasteful and we should be smarter about how and when we call this RPC. Using an exponential backoff instead of a static polling interval might improve performance and UI/UX
Currently z_getoperationstatus is a potentially expensive operation. When a ztx is in the process of creation, z_getoperationstatus will block, waiting until the ztx is submitted to the mempool to return data.
The SD GUI constantly calls
z_getoperationstatusand throws away any errors, which means it's constantly bombarding the full node with these calls if hushd is spending minutes on a complex tx.This whole process is generally wasteful and we should be smarter about how and when we call this RPC. Using an exponential backoff instead of a static polling interval might improve performance and UI/UX