All Collections
EOS Support Media
EOS Node Operator Round Table November 16: P2P improvements (Net Plugin Enhancements)
EOS Node Operator Round Table November 16: P2P improvements (Net Plugin Enhancements)

Published on November 29, 2022

Markus Hinrichs avatar
Written by Markus Hinrichs
Updated over a week ago

Author: Markus Hinrichs

Editor: Randall Roland

The most recent EOS Node Operator Round Table included information about the final release of Antelope Leap 3.2. It also featured the second round of a lively, in-depth discussion about peer-to-peer (P2P) problems and improvement requests. This time concentrating on the improvement of the Net plugin. 14 people attended this meeting.

These weekly meetups are led by EOS Nation's Daniel Keyes and focus on current and upcoming development. They offer useful information for organizations working on software development, such as developers, BPs, blockchain engineers, and members of the Community who wish to learn more about the development process.

By exchanging and interacting frequently, an ecosystem may develop healthily and organically. BPs and developers have given the EOS Network Foundation positive feedback regarding its development. The Community is aware that its voice is now being heard and valued. Last but not least, actions are being done in response to community concerns.

Click here for the video recording of the meeting (Passcode: 8c9R+!P4)

Summary of the Antelope Leap Updates on the way from Stephen Diesel (ENF, Product Manager of Leap)

UPDATES

RELEASE TIMEFRAME

Leap 3.2 final release

within next week

System contract updates

on the way

Release of DUNE

next month (December 2022)

After introducing the upcoming final upgrade to Leap 3.2, Stephen Diesel gave a brief historical background on P2P at EOS. Basically, there have been a lot of complaints before about the P2P NET plugin in terms of stability, performance and the difficulty of finding peers. So that the P2P issue was now given absolute priority by the ENF and the Blockchain Coalition of WAX, TELOS, UX Network and EOS.

Initially, an overly large RFP was created to improve the P2P situation, but it has now been split into several smaller RFPs to make it easier for teams to apply for and work on it and successively improve the P2P situation significantly for block producers, node operators and developers.

Net Plugin Improvements - Round Table Discussion

The main purpose of these Node Operator Meetings is to let those directly affected have their say and to take on board their suggestions, wishes and incentives for improvement and to incorporate them into product development. In the lively discussion between Michael from EOS USA, Ross from EOS Sphere, Kevin from OCI and Matthew Darwin (EOS SysAdmin), some important aspects crystallized, which were summarized in a detailed Github document by Stephen Diesel after the meeting:

Requests to development to make the network more stable and efficient:


  • The Net Plugin, which is optimistic and more or less stateless, should have some intelligence and be able to have at least some state, so to speak, to reduce inefficient workflows and not just blindly run through - this would save valuable time.

  • Improving peer management of nodes. Currently there is little way to evaluate, configure and manage peers in terms of bandwidth usage, total peer connection count. These features could greatly increase the efficiency of P2P data exchange.

  • Improve management over problematic peers that submit unreliable transactions, unlinkable blocks, do not respect the "no transaction" signal, and so forth. Currently there is no easy way to detect and remove them from the peer list.

In essence, the goal is to incrementally enhance P2P without having to full rewrite the protocol, which is currently outside the scope of this project. Given the numerous obstacles in blockchain programming, it is understandable that there might be trade-offs or decisions that have an impact on others. A good example is the trade-off between latency and performance when optimizing throughput by compressing the block production time. But as always, certain groups favor lower latency, such as gamers and others for whom it is not a high issue.

Agenda for the next Node Operator Round Table discussion

For next week's Michael from EOSUSA proposed a discussion on state database trimming. The meeting's host, Daniel Keyes, says that any more discussion ideas are welcome. Comment on this github issue and utilize the reactions to support or oppose the proposals submitted.


Participants (14) of this round table:

  • Randall Roland | EOSsupport.io

  • Dario | EOSsupport.io

  • Kevin Heifner | OCI

  • Michael | EOSUSA

  • Brian Hazzard

  • Jeff Werner | ENF

  • Max Cho | KOREOS

  • Daniel Keyes | EOS Nation

  • Stephen Diesel | ENF

  • Matthew Darwin | EOS Sys Admin

  • Corvin Meyer auf der Heide | liquiid.io

  • Ted Cahall | ENF

  • Ross | EOSphere

  • Bucky Kittinger | ENF


Sources & References

Did this answer your question?