Sui officially released a review of the mainnet downtime incident, stating that the three major failures were caused by upgrades, and no confirmed transactions were rolled back

The official Chinese release from Sui regarding the mainnet downtime incident review states that on May 28, 2026 (Thursday) and May 29, 2026 (Friday), Pacific Time (UTC-7), the Sui mainnet experienced three network failures.

The first two failures were caused by a crash vulnerability arising from the interaction between the Gas billing logic and the recently released version 1.72 (which introduced the Address Balances feature). The fix for the Thursday incident was a temporary solution aimed at quickly restoring network operation while the Sui core team develops a long-term solution. The team was aware that this temporary fix had a very low probability of causing network failures, but accepted this risk to restore the mainnet operation as quickly as possible. On Friday morning, the network triggered another variant of this known issue and failed again.

The third failure occurred during the routine Epoch switch on Friday afternoon when the validator nodes restarted to deploy the fix from Friday morning, triggering a long-standing random number state saving defect, which caused the network to fail again.

The failure times are as follows: First: started around 7:00 PT on Thursday, restored at 13:30 PT; Second: started around 5:00 PT on Friday, restored at 8:30 PT; Third: started around 13:30 PT on Friday, restored at 19:20 PT; throughout the entire event, user funds remained safe, and no confirmed transactions were rolled back after the network recovery. Currently, the validator nodes have fully repaired the original Gas Charging and Randomness State vulnerabilities, and network activity has returned to normal.