In an effort to distribute tokens amongst community members that matter most, validators and miners, we had announced a while back that we will not be conducting a public sale. Instead, the distribution will directly target participants of interest. Two modes of distribution were identified in particular: Eggnet relayers and FlowMint gateways. With the Eggnet designed to incentivize the supply side of the network bootstrapping the decentralized cluster of nodes, FlowMint is to encourage usage and drive bandwidth liquidity in the network.
It brings us great joy to share that the Eggnet was a great success witnessing over 1100 nodes at its peak. It demonstrated the scalability of the network while also proved robustness of comparatively untested code with a vast variety of clients. A number of poorly performing nodes have since quit the network bringing the tally to just over 850 at present.
Some observations:
- Community node operators performed better than several professional staking companies. Infra providers seem to prioritize and have an edge on security. In contrast, Marlin nodes are designed to be run by inexperienced operators without expensive security setups. Non-professional community enthusiasts seemed to display greater hustle in terms of achieving better performance.
- The average time for nodes to hear about Ethereum blocks without Marlin was 351 milliseconds.
- Nodes from locations not initially anticipated like Nigeria, Brazil, Haiti, Australia and Japan participated in the Eggnet.
- The average cost for an operator to run a single relay node came out to be 14 USD per month. We expect this cost to increase a bit as we add support for chains other than Ethereum thus utilizing more bandwidth.
- It took an operator an average of 4.5 days to get the setup and another 1.5 weeks to be competitive.
- Clusters with a node within China and/or in Hong Kong or Taiwan tended to perform better.
Operating a Marlin node is a process of constant learning and experimentation. There are always ways to increase efficiency till the theoretical limits aren't hit. A good set of reference parameters for nodes are documented here:
https://gist.github.com/voluntas/bc54c60aaa7ad6856e6f6a928b79ab6c.
In particular,
- Run
watch netstat -us
- Check if send or receive errors are increasing with time, indicates buffer overflows
- If yes, increase buffer size (replace the numbers below as you wish)
# Per-socket read buffers
net.core.rmem_default = 8192000
net.core.rmem_max = 8192000
# Per-socket write buffers
net.core.wmem_default = 8192000
net.core.wmem_max = 8192000
# Option memory
net.core.optmem_max = 8192000
# Global tuning (multiplied by 4KB)
net.ipv4.udp_mem = 64000 64000 64000
- Restart the relays and check again after some time
It's not just a hope; from our conversations we're certain that validators enjoyed the process of participating in the Eggnet as much as we enjoyed shepherding it. That being said, operating a node for many is an economical for-profit business. And we shalt not disappoint.
Totally, 4.3% of the total supply of Marlin tokens will be awarded to Eggnet participants. If you were an eggnet participant and have completed KYC, you can determine tokens due to you at https://www.marlin.pro/eggnet-rewards. More information about the tokens themselves will be announced in the coming days but feel free to take a sneak peak at https://www.marlin.pro/pond.
The Larvanet, expected to begin within a month, will mark the transition to a more accountable network with nodes required to stake MPOND. Staking rewards in the range of 10-15% have been earmarked for the same. Validators may refer to https://www.marlin.pro/docs/staking/guide/ to begin early preparations to accept external delegations. If you missed the opportunity to participate in the Eggnet, fret not! We hope to see your participation in FlowMint, the upcoming cross-chain stakedrop of Marlin tokens.
In the meantime, follow our official social media channels to get the latest updates as and when they come out!
Twitter | Telegram Announcements | Telegram Chat | Discord | Website