Validators of the Proof-of-Stake blockchain Solana completed a cluster restart at 11 p.m. EST on Saturday after the network was unable to reach consensus for nearly seven hours.
One insider said an “insane amount of data” flooded the proof-of-stake chain, leaving validators unable to reach consensus and grinding production to a halt. On Saturday, bots flooded Candy Machine, an NFT minting tool, with unprecedented traffic: four million transaction requests and 100 gigabits of data per second – a record for the network, according to a Solana Foundation source.
Co-founder Anatoly Yakovenko, who said he was traveling during much of the fracas, credited the validater community for leading the mainnet recovery. He had been criticized on Twitter Saturday for allegedly being “MIA” during a network crisis.
In contrast to last September’s 17-hour outage, Saturday’s hard fork restart failed to bring the new-and-improved code to the validators. Instead, the network continued where it stopped seven hours before.
When preparing for the restart, the validators considered whether they should implement code that temporarily blocked Candy Machine transactions. On Discord, there was a debate about whether such a move constituted censorship. In any case, it would only work if two-thirds of validators agreed, which only a few appeared to do.