BlockBeats news, on March 8th, Lei Yang, co-founder of MegaETH, posted on social media that the RPC crash event yesterday exposed an interesting edge case. Debugging the RPC method calls resulted in the server allocating a large amount of memory for a super long string, ultimately causing memory exhaustion. The MegaETH team decided to build their own RPC stack to apply multiple optimizations on the RPC stack to improve performance. In terms of performance, the effect is very good, but the cost is that the team needs to handle the operation and maintenance of the infrastructure themselves. It is expected that there will be more crash situations in the future, and more post-incident analysis reports will be released. It is hoped that these reports will spark interesting discussions. They encourage everyone to provide more edge cases for testing now, rather than encountering problems months later.
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
MegaETH has decided to build its own RPC stack, and downtime is expected during the testing period
BlockBeats news, on March 8th, Lei Yang, co-founder of MegaETH, posted on social media that the RPC crash event yesterday exposed an interesting edge case. Debugging the RPC method calls resulted in the server allocating a large amount of memory for a super long string, ultimately causing memory exhaustion. The MegaETH team decided to build their own RPC stack to apply multiple optimizations on the RPC stack to improve performance. In terms of performance, the effect is very good, but the cost is that the team needs to handle the operation and maintenance of the infrastructure themselves. It is expected that there will be more crash situations in the future, and more post-incident analysis reports will be released. It is hoped that these reports will spark interesting discussions. They encourage everyone to provide more edge cases for testing now, rather than encountering problems months later.