-
Notifications
You must be signed in to change notification settings - Fork 724
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Improve network parameters #5177
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Strongly in favour
@Mergifyio queue |
🛑 The pull request has been removed from the queue
|
@Mergifyio requeue |
✅ This pull request will be re-embarked automaticallyThe followup |
✅ The pull request has been merged automaticallyThe pull request has been merged automatically at 4db84de |
* Modify network parameters for current mainnet conditions
The current state of mainnet is changing due to how clients are subscribing to long lived subnets. The modification has lead to more scarcity in the subnets as we expected.
We are now seeing issues where clients are unable to find and maintain useful peers to span all subnets, leading to missed attestations and the warning log, failed to publish message with the error
InsufficientPeers
. It is not enough for individual nodes to simply increase their peer counts, because it is often the case that other nodes on the network have already reached their peer limits and non-lighthouse clients often simply reject the connection. The best solution currently (and has been planned for a while) is to increase lighthouse's defaulttarget-peer
count to ~120. However, we want to do this gradually. This PR takes an intermediate approach and sets the default value to 100.We have been testing a default value of 100 peers for a while now and load seems acceptable for small home users. See attached:
There is a secondary issue where Lighthouse attempts to maintain 6 peers for every required subnet. For nodes with lots of validators this becomes an impossible task with an 80 peer target and results in continuous discovery queries.
With a target peer count of 100, and each peers subscribing to two subnets, a healthy target peer count per subnet is 3. This should result in little to no discovery queries once the right set of peers are found. Having all nodes on the network update to this logic, should allow for more connections between all peers and more stability on the subnets.