• add how far behind (latest block height) syncing check amongst healthy clients
• use priority queue to store most up-to-date client as the highest priority
• also allow optionally choosing a sticky connection to a configured client

Retry/re-create the subscription in the job_manager for subscribing jobs to the after some time of a stale subscription. Useful when for ex: eth subscription to certain filtered logs, but no logs for a long time, "slow subscribers" will get dropped by the RPC node.

On restart, should check mempool to see if any tx is stuck as pending, in which case the next tx should be sent on the same nonce with a gas bump.

Add detailed comments to each function describing its purpose, parameters, return values, and potential side effects. This enhances readability and helps other developers understand how to interact with the interface.

so we can plug in data dog but don't kill prometheus

Implement w3 for a BatchClient that supports easy batch calls. Will reduce network overhead for applications that make eth_calls, etc. often

Currently, in the case where a node’s mempool has queued txs for the account (on-startup) but no pending txs, the next tx is sent on top of the queued txs

Solution: Send the next tx with the pending nonce, rather than following the queued,

artio.bex.berachain. t

ransactions freeze during exchange from Rabbit wallet/

If switching between, multiple RPC endpoints, will break invariants held by Transactor.