Testnet4 is unreliable

[NotATether]

When testnet4 was released, I was happy just like everyone else around here, but now that I'm trying to test something I've developed with Electrum addresses in testnet mode, I actually need to use testnet4 for something serious.

Faucets and acquiring bitcoins are not the issue. The issue is, strangely enough, that my faucet transactions never get confirmed, because nobody is mining them.

Everybody just seems to be mining empty blocks and there's a small chance that someone sweeps up the few transactions in there to confirm them.

What's the point of testnet if nobody can actually test stuff? Argh.

Might consider pointing my rented RTX 3050 at some pool with a GPU miner software and mine myself if this continues.

I know about signet, don't tell me to use it though because Electrum doesn't support it.

[Antidote47k]

That sounds frustrating, from what I’ve been reading I think aside from the lack of miners, the 20mins minimum difficulty rule on testnet4 may have unintentionally created an incentive for  miners to race for difficulty 1 blocks. Since an empty block is faster to construct and propagate, many simply mine empty blocks which leaves transactions sitting in the mempool. This seems to defeat the essential purpose of a test network where developers actually need transaction confirmations to test their software.

There are discussions about addressing this behavior cause it has become a usability problem. If developers have to resort to mining their own blocks just to get transactions confirmed then that is probably a sign that something in the incentives isn’t working as intended

https://delvingbitcoin.org/t/disabling-min-difficulty-blocks-on-testnet4/2350

[NotATether]

This seems to defeat the essential purpose of a test network where developers actually need transaction confirmations to test their software.

Exactly. And I can't believe I'm saying this, but I'm actually starting to miss testnet3. That one was reliable at least, even if coins became so scarce that people were resorting to selling them.

Like I said, I would've simply used Signet and side-stepped all this nonsense, but it is not supported by Electrum which is part of the development I'm trying to test.

Edit: three transactions are now not going through and instead just get dropped from the mempool after some time. WTF?

[athanred]

What's the point of testnet if nobody can actually test stuff?

The same, as it is for many other altcoins: to mine it, and sell for BTCs: https://altquick.com/exchange/market/BitcoinTestnet4

They listed even signet: https://altquick.com/exchange/market/BitcoinSignet

Which means, that the time, when testnets were worthless, is gone. Use regtest, if you really want to test things. They won't list a coin, where six blocks per second can be produced.

Might consider pointing my rented RTX 3050 at some pool with a GPU miner software and mine myself if this continues.

Then try testnet5: https://github.com/bitcoin/bips/blob/master/bip-0095.md

Connected topic: https://bitcointalk.org/index.php?topic=5586192.0

I know about signet, don't tell me to use it though because Electrum doesn't support it.

You can use even mainnet as a testnet, if you really need it. Your Electrum client will trust the data you feed it with. It is only an SPV client: you don't even need a real chain to make things tick; it would work as well in a pseudo-chain setup, with valid merkle trees. Also, if you start a fake mainnet chain, and mine things on top of block zero, then everything would work as well.

I'm actually starting to miss testnet3

Nobody stops you from using it. You can even use zero satoshis to test things, because why not. Everything is possible, if you can detach your client from the original test network, and mine a valid block on your CPU, after 20 minutes, no matter if you use testnet3 or testnet4. When you reconnect, all of these coins will be reorged, but well: do you want to test things, or do you want to have coins, which are accepted by other people? By being publicly accessible, all test coins gained non-zero value. Use regtest or a custom network, if you want to have truly worthless test coins.

[PVequalNRT]

Might consider pointing my rented RTX 3050 at some pool with a GPU miner software and mine myself if this continues.

I know about signet, don't tell me to use it though because Electrum doesn't support it.

I don't think this the permanent solution for the chain, the testnet3 was broken because of the way way miners takes advantage of hashrate drop and now Testnet4 does have it's challenges now. May be this is to tell us that testnet cannot be as effective as mainet even if there is another testnet in the future.

I was looking for testnet4 tokens some days ago, I received some sats from faucet but I needed more. I message a verified vendor on Telegram and asked how much per tbitcoin, was given $1 per token, I never knew miners has incentivise testnet4 and this expensive. If the mining is not decentralized, miners will benefits and careless about the network.

[NotATether]

I was looking for testnet4 tokens some days ago, I received some sats from faucet but I needed more. I message a verified vendor on Telegram and asked how much per tbitcoin, was given $1 per token, I never knew miners has incentivise testnet4 and this expensive. If the mining is not decentralized, miners will benefits and careless about the network.

How are you even supposed to receive said tbitcoin if it takes an unknown, indefinite time to get the transaction in a block?

Anyways, this is what I was developing. I ended up using mainnet to test it and spent $5, and I couldn't test the parts where change addresses are supposed to be created even though it should use Electrum's function, because Electrum sweeps such small test transactions into the tx fee instead of making dust change.

[LoyceV]

Might consider pointing my rented RTX 3050 at some pool with a GPU miner software and mine myself if this continues.

That won't work: testnet difficulty is far too high for GPU mining, unless you're fast enough to beat a low difficulty empty block miner.

Exactly. And I can't believe I'm saying this, but I'm actually starting to miss testnet3.

Here's an idea: let's just restart testnet3 from scratch. Okay, seriously, let's not go down there, the block storms would quickly mine most of the coins and it's back to the current state.

How are you even supposed to receive said tbitcoin if it takes an unknown, indefinite time to get the transaction in a block?

I think I've seen 8 or more hours with only empty (easy) blocks being mined, until an ASIC miner creates a (difficult) block and includes the transactions.

[athanred]

Here's an idea: let's just restart testnet3 from scratch.

You know, that anyone can start testnet5, right? The BIP is there, so it is just a matter of implementing it. Then, by talking to any block explorer like mempool.space, you can make it visible for users, and then, it can be later included into Bitcoin Core, if it would work.

Testnet4 was included, when around 40k blocks were already mined. And around block 30k, many developers used it, by building it from the source code. The same can happen with testnet5: the first working implementation will get included, so anyone has the same chances of making it, as BIP authors. The only criteria for inclusion is if it would work correctly, and be useful for testing.

So, if you are a developer, and you want to use a working testnet, then you can launch testnet5 as well, and then create a Pull Request into Bitcoin Core repository. If you don't want to do that, then you are stuck with existing testnets, for better or worse.

[LoyceV]

You know, that anyone can start testnet5, right?

If it would be that easy, someone would have done it already.

So, if you are a developer, and you want to use a working testnet, then you can launch testnet5 as well

Last time I tested an exchange, it was already hard to get some test coins. Developers want to spend their time on their own project, not on creating their own testnet.

[athanred]

If it would be that easy, someone would have done it already.

You could be surprised, how many low-hanging-fruits are there. For example: the only thing, needed to mine testnet4, was to just build the source code, and then you could have thousands of coins, mined on CPU. And many people didn't want to do just that, because of lack of skills, or fear, that they will burn a lot of electricity to mine something, which won't be included into Bitcoin Core.

Developers want to spend their time on their own project, not on creating their own testnet.

If someone is smart enough to develop a complex project, then that person should be also smart enough, to change the Genesis Block to anything else, and make a new chain. Altcoin developers do that constantly, and there are more than enough examples, how to start yet another chain.

Anyway, if someone is trying to develop new things, then they should be compatible with BIP-54. So, using testnet5 is recommended anyway, because if some new feature will be incompatible with that, and will for example require making 64-byte transactions, then it will have low chances of being deployed.

Yet another thing is that signet or regtest can be used to test almost everything. There are really few cases, where those two networks are insufficient. And if some software supports freshly deployed testnet4 from 2024, and doesn't support signet from 2020, then something is really messed up.

[ABCbits]

I know about signet, don't tell me to use it though because Electrum doesn't support it.

Do you actually mean it doesn't work properly or bugged? I checked Electrum release note and found it was supported since version 4.1.5 (middle 2021).

# Release 4.1.5 - (July 19, 2021)
--snip--
 * add support for signet test network (#7282)

There's no mention signet support was removed either.

If it would be that easy, someone would have done it already.

You could be surprised, how many low-hanging-fruits are there. For example: the only thing, needed to mine testnet4, was to just build the source code, and then you could have thousands of coins, mined on CPU. And many people didn't want to do just that, because of lack of skills, or fear, that they will burn a lot of electricity to mine something, which won't be included into Bitcoin Core.

Until you deal with small details that may need to be checked or changed. For example, i know Electrum connect to multiple server to get block header data as verification.

By default, Electrum tries to maintain connections to ~10 servers. The client subscribes to block header notifications to all of these, and also periodically polls each for dynamic fee estimates. For all connected servers except one, that is all they are used for. Getting block headers from multiple sources is useful to detect lagging servers, chain splits, and forks.

But if i make Electrum connect to Electrum server that use my custom testnet (which have different genesis block and block header), i don't know whether i need to change Electurm source code or not.

P.S. I truly don't know the answer, someone with actual experience are welcome to answer it.