Evaluate whether we can consider a couple of Subxt errors recoverable, as they are currently forcing a restart of the miner.

[sigurpol]

Disclaimer

Restart of the miner are not critical since the miner is designed to properly recover even if restart happens in the middle of a submission**

Critical errors seen on the field on WAH

We have seen on WAH the miner restarting because of

"Invalid block hash"

025-09-05 15:02:11.667	
Error: Subxt(Rpc(ClientError(User(UserError { code: -32801, message: "Invalid block hash", data: None }))))
	
	
2025-09-05 15:02:11.664	
2025-09-05T13:02:11.664235Z DEBUG polkadot-staking-miner: round of execution finished. outcome = Err(Subxt(Rpc(ClientError(User(UserError { code: -32801, message: "Invalid block hash", data: None })))))    
	
	
2025-09-05 15:02:11.664	
2025-09-05T13:02:11.664140Z ERROR polkadot-staking-miner: Runtime upgrade subscription error: Rpc(ClientError(User(UserError { code: -32801, message: "Invalid block hash", data: None })))    

"Not enough data to fill buffer"

	
2025-09-10 10:04:58.658	
2025-09-10T08:04:58.658617Z TRACE polkadot-staking-miner: Listener task started, watching for finalized blocks    
	
	
2025-09-10 10:04:57.646	
Error: Codec(Error { cause: None, desc: "Not enough data to fill buffer" })
	
	
2025-09-10 10:04:57.638	
2025-09-10T08:04:57.638101Z DEBUG polkadot-staking-miner: round of execution finished. outcome = Err(Codec(Error { cause: None, desc: "Not enough data to fill buffer" }))    
	
	
2025-09-10 10:04:57.638	
2025-09-10T08:04:57.638073Z ERROR polkadot-staking-miner: Miner task failed: Codec error: `Not enough data to fill buffer`    
	
	
2025-09-10 10:04:57.634	
2025-09-10T08:04:57.634144Z TRACE polkadot-staking-miner: get_block_state: Fetching round    
	
	
2025-09-10 10:04:57.634	
2025-09-10T08:04:57.634120Z ERROR polkadot-staking-miner: Critical miner error - process will exit: Codec(Error { cause: None, desc: "Not enough data to fill buffer" })    

The error happens when fetching the round from storage.

let current_round = storage
		.fetch_or_default(&runtime::storage().multi_block_election().round())
		.await?;

To recover or not to recover

Evaluate if recovering instead of restarting for these errors.