Filecoin testnet launched, official first AMA fully recorded

From 10:00 to 11:00 am on March 6, 2020, Beijing time, Juan, the founder of Filecoin, will hold an online AMA to answer community questions. As soon as the news came out, communities large and small became extremely lively.

Interstellar Vision has compiled all the contents of the first AMA when the first phase of the Filecoin test network was launched on December 12 last year, so that everyone can "review the past and learn the new." At the same time, we are also looking forward to the second AMA tomorrow. Interstellar Vision will broadcast the entire process in real time in the community to let more friends understand the progress of Filecoin.

The first AMA details are as follows:

I think it’s very interesting that they have written custom mining software to participate in the network, it shows that we need to improve the built-in Lotus tools, and I also expect to see many other organizations develop custom mining software soon!

2

Will network latency be a problem for Chinese miners? Since China’s internet connection is a bottleneck for the global internet.

We recommend using a lower latency network connection to submit PoSt on time. If a miner fails to submit PoSt on time multiple times, it may result in winning less block rewards and increase the risk of being slashed.

3

In this test network, the power of one company exceeded 51%. In this case, will other miners be able to participate in the later stages of mining?

It looks like one company is getting more ready for a testnet launch than any other. They haven't shown this advantage, so new miners should be able to join and gain power. If they end up not doing well (which I really doubt they will), we may have to reset the network.

4

Will FIL mining become a GPU competition in the future? Are the nodes currently ranked at the top because of their powerful GPUs?

Filecoin mining is not GPU mining. The Filecoin network rewards miners based on their storage computing power.

Filecoin mining is based on storage computing power consensus: having more computing power (proven storage capacity) on the network is directly correlated to getting the answer: more block rewards.

The computing power of a miner on the network is related to the storage capacity of the miner on the network, not proportional to the number of GPUs it runs.

5

What if miners found a mining algorithm that could contribute significantly more rewards than resources on testnet/mainnet?

The security of the protocol is Filecoin’s number one priority. Our protocol design and proof teams have invested heavily in ensuring that the best way to earn mining rewards is to provide a large number of proven hard drives. We’ll continue this work to ensure miners are always incentivized to provide useful services to the network. If you discover a great mining strategy, we’d love to hear about it on Slack!

6

“Outsource SNARK computation to third-party service providers.” How to avoid centralization of these third-party organizations?

1) Make it very easy to provide the service - if someone is malicious or overcharging, someone else can provide it too.

2) Encourage the existence of many (30+) different service providers

3) You can always run it yourself - if it costs more to use a service than to do it yourself, then you'll do it yourself. - This means that service providers have to charge fairly, otherwise no one will use them.

7

The top three & t01050/t01116/t01081 are the same company, not good news. Is it normal?

The testnet is a testing network that was launched less than 24 hours ago. It is still under active development and we expect the network to become more stable as we continue implementation and the number of miners grows. As the testnet grows, it will get closer to more realistic network conditions.

8

If I save a file in a miner, the corresponding sectors have not been filled yet. How does this affect retrieving my file?

Since there are transactions on-chain, users can save files before miners seal them into sectors. Details can be found in the Filecoin spec (https://github.com/filecoin-project/specs). Then, when miners have enough transactions, the sector can be sealed and users can retrieve files from it later. Note that users can also retrieve their files immediately before they are sealed into sectors. This is not currently implemented, but we plan to do it soon.

9

Any suggestions on how miners should handle multiple disks in their miners? RAID, ZFS, etc?

I thought about it, it depends on the amount of storage collateral you want to provide for the data you store, if it's a large amount I would go with a RAID setup (personally prefer RAID10 as repair time on RAID5/6 is ok).

However, if you are a "cheap" storage provider with low prices and little collateral, it might make sense to store everything in JBOD and bear the cost of failure. Whatever you do, don't use RAID0 as a single disk failure will cost you all your storage.

10

It has been more than 12 hours since the testnet was launched, why are there only 4 miners left on the list?

Currently, there are more than 260 miners registered on the network, and the storage startup process is a bit slow, with only miners with more than 512GB of storage power showing up. In fact, we didn’t even expect so many miners to enter the leaderboard so quickly!

11

In the latest roadmap, you said that the mainnet might be launched between March 23rd and April 24th, 2020. I was wondering what are your criteria for deciding to launch it?

The most critical requirement for launching Mainnet is that Filecoin is secure. We will not launch an insecure network. The next requirement is performance and functionality. Currently, the launch window is our best estimate. As we get closer, the window size will shrink and we will be able to pick a date.

12

Will there be rewards in the second phase of the testnet?

Moderator: We don't have any rewards for this at the moment.

13

If the entire city was without power for 24 hours, how would you punish storage miners?

Details about failure and recovery can be found in the specification (https://github.com/filecoin-project/specs/). The protocol will meet the practical operational needs of good miners and service providers, while maintaining the security of the protocol and the quality of service for customer data. Taking data storage services offline is not good for the client, nor for the service. At the same time, temporary power failures will occur, and the protocol needs to be resilient to these failures to some extent. As we get closer to mainnet, we will set cryptoecon parameters and policies and explain them clearly for everyone.

14

Is there a non-linear relationship between penalty and disconnection time?

As we get closer to mainnet launch, it’s possible that many answers to specific cryptoeconomics questions will be published during the testnet. Stay tuned. In the meantime, you can view the list of cryptocurrency placeholders in the Filecoin Spec here:

https://filecoin-project.github.io/specs/#algorithms__cryptoecon

15

If a miner uses the "lotus-storage-miner pledge-sector" command, how long will the storage task last? How to avoid the pledge being lost in the production chain?

We're going to make it customizable, set it to a pretty big value right now in Lotus. We might even build a special mode for "filling storage" where you add storage just to increase the charge without any real transaction.

16

Filecoin testnet hardware parameters?

You can learn more about the standard test configuration that Protocol Labs uses as a testnet on the blog at https://filecoin.io/blog/filecoin-testnet-mining/. However, this configuration is not guaranteed to be suitable for Filecoin storage mining at mainnet launch and is only used for testing.

17

Can we attract/host data storage needs from real projects during testing?

The network may experience resets during the testnet. It is not recommended to store customer data or other data that requires very reliable storage before the mainnet.

18

First of all, congratulations on the launch of the Filecoin testnet. I wonder under what circumstances, as you said in your latest roadmap, the testnet will be restarted.

In most cases we try to fix all bugs, however, if the problem requires non-trivial changes that the development team must implement, a reset may be necessary. Basically, it will be a trade-off between development costs and network reset costs.

19

It seems that the testnet is quite slow, what is your performance standard for the testnet and in what ways can you improve it?

The software used for the testnet has not yet been optimized for performance, the focus is on making sure everything works properly and the network performs its primary functions. We still have a lot of work to do in terms of optimization.

20

When will the official wallet be developed?

We are working with several hardware and software wallets, and the official wallet should be released before the mainnet goes live.

twenty one

Will we have an official Filecoin wallet?

We are working with multiple hardware and software wallet projects and hope to have reliable Filecoin wallet support before mainnet launch.

There is also a developer grant RFP for wallets https://github.com/filecoin-project/devgrants/blob/master/rfps/rfp-filecoin-wallets.md that expires on January 15th.

twenty two

The recommended hardware is a huge barrier to entry for miners. Is the design decision worth considering, or you would not have a home miner with a small home server and NAS?

We are working hard to reduce the cost of sealing and proving. We do want to support smaller miners, but not at the expense of security. It is always possible to make things faster and cheaper through future optimizations, but it is difficult to make things more secure.

twenty three

You consider adding outsourced GPU computing. Isn't this a centralization risk? Will access to the network be impacted during outages of these central providers?

This should not be a significant centralization risk. The cost of buying your own GPU or launching a new outsourced verification service will be very low. Even if many people decide to use third-party attestation, there should be a competitive and well-distributed market.

twenty four

Are the hardware requirements for participating in the FIL test network too high?

We recently announced our standard testing configuration here: https://filecoin.io/blog/filecoin-testnet-mining/. This configuration will change over time as we optimize network parameters. In our blog post, we highlighted “specific tradeoffs related to computing power made in this standard testing configuration” by allocating resources to GPUs rather than adding more CPU power.

It is possible to mine on the testnet with other hardware configurations, but we have not tested all of them. We expect testnet miners to find other hardware combinations, some of which may be more efficient than the test configuration described in the blog post. If you use a different hardware configuration to run Lotus, please run the Lotus benchmark tool and post the results in this thread (https://github.com/filecoin-project/lotus/issues/839).

25

Does filecoin in testnet really work?

The testnet is now live! You can check the status at https://stats.testnet.filecoin.io and learn how to join the network using the Lotus client at https://docs.lotu.sh.

Please note that Filecoin tokens on the testnet are only used for testing and have no value.

26

The first miner collateral program will be announced in March?

Collateral is very important. They ensure correct and secure operation in the network. Filecoin storage miners must post Filecoin as collateral to mine blocks (“Collateral”) or participate in storage transactions (“Storage Collateral”). We will provide more information about how staking and trading collateral works in the Filecoin specification before mainnet (please note the latest roadmap:

https://filecoin.io/blog/roadmap-update-2019-q4/). You can see the list of cryptoeconomics placeholders in the Filecoin spec here:

https://filecoin-project.github.io/specs/#algorithms__cryptoecon

27

Can't use pacman opencl-icd-loader package on entos7?

This is an implementation-related troubleshooting question. Please ask again in the #fil-lotus channel on Slack.

28

So can we ask troubleshooting questions here? Or fil-help file-lotus?

For questions about troubleshooting the implementation, please again ask in the #fil-lotus channel on Slack.

29

I have a power of 0: Power: 0 B / 19.2 TiB (0.0000%) Is this normal?

You have to seal sectors of 512GB or more for it to show any functionality.

30

Network latency and bandwidth, which is more important in mining?

Definitely make sure you have low latency (ideally, less than 100ms to other major countries), but I recommend at least a 50 megabit internet connection for bandwidth.

31

Is it reasonable that using "sealed workers" will lead to some unnecessary competition and make it difficult for ordinary miners to compete?

Sealing workers are used to distribute sealing work to different machines. Unless we can make sealing really cheap, this will always be impossible (although at some point you will max out the write speed of your hard drives). We provide these tools to help level the playing field for miners who will write their own versions of the software anyway.

32

Data storage in Filecoin requires PoSt and other proofs. How long does it take to store 1GB of data in Filecoin?

As a client, you should be able to store gigabytes of data as fast as the miners can transact with you (an automated process) and transfer the data to them.

33

Why are no miners with only 32GiB power (which is the standard in faucets) showing up on the dashboard? How did those miners with more than 1TiB join?

Before miners appear in the power table, our minimum miner size is 512 GiB and can receive block rewards. You can achieve 512 GiB with 1GB or 32GB sectors. Miners with less than 32GB will not receive block rewards, but can still accept storage deals. Please note that these are testnet parameters and may change before mainnet.

34

Is there clear information about the hardware requirements?

, we recently published a blog post discussing our hardware testing configuration here: https://filecoin.io/blog/filecoin-testnet-mining/. Please note that there are many different hardware configurations you can use to mine the Filecoin testnet. Our standard testing configuration describes the hardware our team tests on. As the testnet evolves, our standard testing configuration will change over time as we refine the parameters of the network.

35

Is a public IP always a requirement for mining Filecoin?

Yes, you need a public IP address, or you can have a separate machine with a public IP address that relays the connection back to the internal machine. The requirement is that you must be dialable on libp2p.

36

As a new generation of proof of storage blockchain, how do you evaluate or ensure the consensus security of the current testnet? For example, VDF's ePost?

We performed an extensive security analysis of all parts of the protocol. We worked through a variety of attacks - following the work on the https://github.com/filecoin-project/specs repo and the Filecoin Research https://research.filecoin.io/ website.

37

What are the currently recommended specs for testnet mining that are most likely to make the hardware also useful for mainnet mining?

Hello, we recently published a blog post discussing our hardware testing configuration here: https://filecoin.io/blog/filecoin-testnet-mining/. Please note that there are many different hardware configurations you can use to mine the Filecoin testnet. Our standard testing configuration describes the hardware our team tests on. This configuration may change over time as we refine network parameters as the testnet evolves.

38

How do transactions and payment channels work and is the entire process automated?

Filecoins payment channels are similar to the various payment channel constructions implemented on Ethereum. Their code is implemented and works, but there is nothing automated around us.

39

Will miners be penalized if they don’t accept a transaction that meets their latest asking price?

Transaction negotiation is done off-chain and miners will not be penalized at the protocol level. However, services and platforms can be built on top of the protocol and may incur some penalties.

40

Miners should prove that it has free storage available?

This is actually part of the key economic mechanisms we are considering for the protocol. The Filecoin specification lists some other mechanisms and parameters that may still change during the testnet, but must be resolved before the mainnet launch. You can see the list of cryptoeconomic placeholders in the Filecoin specification here: https://filecoin-project.github.io/specs/#algorithms__cryptoecon

41

If mainnet is 100, what would your score be for testnet?

42 points.

42

How much money can you make as a miner?

After mainnet launch, when real Filecoin tokens will be available, this will depend on many things, including how many others choose to mind and the extent and speed of Filecoin adoption. We will publish more information about Filecoin’s cryptoeconomic model in Q1, which may help you refine your own calculations.

43

Is there a scenario where you might lose your files?

No matter what type of storage provider you use, there is always a chance that a hard drive will fail or go offline. Today, centralized providers solve this problem by storing multiple redundant copies of your data. Filecoin gives you control to do the exact same thing in a truly decentralized way by replicating your data with multiple miners, based on your cost and reliability preferences.

44

If GPU computing is not considered a centralization risk. How likely is it that specific ASICs could be created to efficiently compute and centralize proofs?

Filecoin mining is not GPU mining. Filecoin rewards miners based on the proportion of storage they add to the network. Filecoin mining is based on storage power consensus: having more power (proven storage) on the network is directly correlated to winning more block rewards. The amount of power a miner uses on the network is linearly related to the amount of storage a miner has on the network, not the number of GPUs they are running. The Filecoin testnet can be mined using many different hardware configurations.

- over-