Adding Support for StratumV2 Protocol

[chGoodchild]

Enhancement Request: I would like to propose adding support for the StratumV2 protocol to the ESP-Miner project. StratumV2 offers significant improvements over the current StratumV1, including enhanced security, reduced bandwidth requirements, and the ability for miners to construct their own block templates.

Key Benefits:

• Security: StratumV2 provides better protection against security threats, implementing end-to-end encryption to prevent hashrate hijacking and other man-in-the-middle attacks Stratum V2 Overview.
• Efficiency: The protocol reduces unnecessary data transmission, optimizing operations by using a more efficient binary encoding, which significantly lowers the data payload size compared to the JSON-RPC used in Stratum V1 Stratum V2 Documentation.
• Decentralization: Allows greater control over block construction, supporting decentralized mining practices by enabling miners to select their own transaction sets, thus improving the decentralization of mining Bitcoin Manual on Stratum V2.

Consideration: While it is possible to use a translation proxy to avoid making significant changes to the ESP-Miner, this approach would not allow hashers to benefit from the direct security and efficiency improvements offered by StratumV2. Thus, implementing StratumV2 natively in ESP-Miner could provide more substantial benefits to mining operations.

Thank you for considering this addition.

[chGoodchild]

Sjors already created bitcoin core binaries for his StratumV2 pull request. There is a testing guide here.

[benjamin-wilson]

I would suggest it does not improve decentralization as the pool can always reject a template proposal and or not propagate a valid solution.

I agree on the security and efficiency, however both of those benefits can be gained by using the translation proxy outside of the LAN.

IMO esp-miner could benefit but this is a low priority item.

[chGoodchild]

Perhaps I can try to implement it once I have built a Bitaxe that I can test with. Would you be open to PRs in this direction?

[benjamin-wilson]

Absolutely

[TheBlueMatt]

I would suggest it does not improve decentralization as the pool can always reject a template proposal and or not propagate a valid solution.

This is a common misconception and inaccurate - the Stratumv2 reference implementation (SRI) and the intended use mandate that if the pool declares they wish to reject your template that you immediately fall back to an alternative pool or solo mine. There is no world where an Sv2 mining device which is doing custom template selection will ever stop mining on the template it wants to, the pool saying no just may change how the payouts for that job work.

[skot]

The Bitaxe controller (ESP32) is a pretty resource constrained device.. do you think it would be better to have a SV2 proxy running on your node that you point your SV1 Bitaxe to?

[TheBlueMatt]

For those doing custom work selection there's no much difference, indeed. However, for those doing legacy pooling Sv2 would at least substantially improve their security. ISTM its worth at least seeing if a minimal Sv2 implementation can fit on the ESP32, but if not its not a huge deal.

[Sjors]

In sv2 block template creation is handled by the node, which presumably is running on a different machine. I think there's three approaches here:

1. On the device with the full node, run the SRI Job Declarator Client and the Translator roles. The miner on the device simply points to the translator and can keep speaking sv1. Main downside: the translator adds complexity, more stuff to configure, makes debugging more difficult, etc. But it's the easiest place to start.

2. Make the ESP32 firmware speak native Sv2, i.e. add noise encryption and small number of messages. Then on the node machine you only need to run the SRI Job Declarator Client.

3. Add the Job Declarator Client to the ESP32 firmware. It will fetch job templates from the full node, and communicate with the pool. This requires much more data processing on the ESP32 device, but not block validation so maybe it's doable. The major benefit is that it's plug and play. User just needs to provide a pool address (and job declarator server address if it's different).

Though perhaps in some future upgrade Bitcoin Core could also handle the work done by the Job Declarator Client. The current PR doesn't.

[benjamin-wilson]

I would suggest it does not improve decentralization as the pool can always reject a template proposal and or not propagate a valid solution.

This is a common misconception and inaccurate - the Stratumv2 reference implementation (SRI) and the intended use mandate that if the pool declares they wish to reject your template that you immediately fall back to an alternative pool or solo mine. There is no world where an Sv2 mining device which is doing custom template selection will ever stop mining on the template it wants to, the pool saying no just may change how the payouts for that job work.

I never suggested the device would stop mining but the pool can effectively censor the transactions you proposed, you must go somewhere else, similar to v1. Or worse they accept the template but withhold a valid block.

Add the Job Declarator Client to the ESP32 firmware. It will fetch job templates from the full node, and communicate with the pool. This requires much more data processing on the ESP32 device, but not block validation so maybe it's doable. The major benefit is that it's plug and play. User just needs to provide a pool address (and job declarator server address if it's different).

3. is almost certainly too much for the esp to handle on top of the mining work.

[TheBlueMatt]

I never suggested the device would stop mining but the pool can effectively censor the transactions you proposed, you must go somewhere else, similar to v1. Or worse they accept the template but withhold a valid block.

The key value here is that it changes an active action to a passive one. We've seen time and time and time and time again that miners don't really care and will simply do whatever their software does for them. In the current world, if a pool starts censoring, miners learn nothing about it, they keep on happily mining the censored transaction, and we rely on one person (0xB10C is, as far as I'm aware, the only one monitoring to detect such things) to notice this is happening and then get the word out via Twitter to miners and then hope that they tear up their existing business contracts and move to a new pool. During all this time Bitcoin is being actively censored.

Instead, with Sv2, miners selecting their own work will continue to mine the work that contains any transaction the pool wants to censor, and then one of two things will happen. Either (a) the pool will send a rejection message to the Sv2 miner who's software will then automatically fall back to a fallback pool or solo mine or (b) the pool will do nothing and simply not pay out, at which point the miner can go to their existing contract and either sue the pool for breach and lost revenue or move to another pool and throw away a few days of revenue.

Sure, a miner could, when they notice the issue, go back to their previous pool and start mining censored blocks, but that takes an active decision on the part of the miner some substantial (human-response) time after some censored transaction(s) were mined. This is an absolutely massive change compared to today.

[benjamin-wilson]

I never suggested the device would stop mining but the pool can effectively censor the transactions you proposed, you must go somewhere else, similar to v1. Or worse they accept the template but withhold a valid block.

The key value here is that it changes an active action to a passive one. We've seen time and time and time and time again that miners don't really care and will simply do whatever their software does for them. In the current world, if a pool starts censoring, miners learn nothing about it, they keep on happily mining the censored transaction, and we rely on one person (0xB10C is, as far as I'm aware, the only one monitoring to detect such things) to notice this is happening and then get the word out via Twitter to miners and then hope that they tear up their existing business contracts and move to a new pool. During all this time Bitcoin is being actively censored.

Instead, with Sv2, miners selecting their own work will continue to mine the work that contains any transaction the pool wants to censor, and then one of two things will happen. Either (a) the pool will send a rejection message to the Sv2 miner who's software will then automatically fall back to a fallback pool or solo mine or (b) the pool will do nothing and simply not pay out, at which point the miner can go to their existing contract and either sue the pool for breach and lost revenue or move to another pool and throw away a few days of revenue.

Sure, a miner could, when they notice the issue, go back to their previous pool and start mining censored blocks, but that takes an active decision on the part of the miner some substantial (human-response) time after some censored transaction(s) were mined. This is an absolutely massive change compared to today.

Do you have any insight on the possible block withholding?

[TheBlueMatt]

What possible block witholding, not sure what you're referring to. A miner doing custom JD will broadcast any blocks they find directly (in addition to the pool doing so).

[benjamin-wilson]

A miner doing custom JD will broadcast any blocks they find directly

Ok, that's what I was looking for. I was unsure if this was expected default behavior.

[plebhash]

I don't think it makes sense to integrate Job Declaration Protocol (negotiating templates) into the ESPminer.

This part of the protocol is not relevant at a firmware level.

If the miner wants to participate in Job Declaration, they should deploy a Job Declaration Client and one or more Proxies on their LAN infrastructure.

The ESPminer simply sends shares to the Proxy, which forwards them to JDC.

With SV2, the ESPminer will be consuming less bandwidth and computation, so it's still an improvement.

[Sjors]

If the miner wants to participate in Job Declaration, they should deploy a Job Declaration Client and one or more Proxies on their LAN infrastructure.

This seems too complicated, unless you have some node-in-the-box solution that installs and configures all that.

is almost certainly too much for the esp to handle on top of the mining work.

Right, so this needs to be something that's easy to install next to the node.

[plebhash]

After some reflection, I'm redacting my previous comment where I said that it doesn't make sense to add JD functionality to ESPminer.

Assuming the ESP32 has enough computation power, it would be an interesting feature (as @Sjors already described above).

[plebhash]

@benjamin-wilson already stated that this is a low-prio task, so it will probably take a while before it starts being implemented. Although it seems maybe @chGoodchild might eventually give it a shot?

Anyways, if you guys eventually start planning to dive into this, please let the SRI contributors know on the github issue linked above, because FFI coverage will have to be expanded. Unless you want to do the full-Rust avenue, or you use some pure-C implementation (which I'm unaware of, so please let me know if you find some).

[benjamin-wilson]

A rust implantation of esp-miner is in the very early stages of development. I'm not sure how much it will be worked on.

https://github.com/bitaxeorg/esp-miner-rs

[chGoodchild]

@benjamin-wilson already stated that this is a low-prio task, so it will probably take a while before it starts being implemented. Although it seems maybe @chGoodchild might eventually give it a shot?

Anyways, if you guys eventually start planning to dive into this, please let the SRI contributors know on the github issue linked above, because FFI coverage will have to be expanded. Unless you want to do the full-Rust avenue, or you use some pure-C implementation (which I'm unaware of, so please let me know if you find some).

I haven't started yet. This is something I really want to work on, but realistically I won't be able to start before July.

I'll be sure to ping you and other SRI contributors before starting to make sure I have a viable/suitable approach.

Thanks!