On May 31, the Sui Foundation acknowledged in a post-mortem that the three network disruptions that occurred between May 28 and 29 stemmed from a new feature (address balance for paying fees) introduced in version 1.72 released eight days earlier, and that the team deployed one of the fixes knowing that the network could go out again.
The first arrests began at 14:00 UTC on Thursday and continued until 20:30 UTC. To resolve it, the team deployed an interim fix. According to the report itself: “Known issues unlikely to result in arrest.”. The team “took a risk” to reactivate the network while developing a more robust solution. At 12:00 UTC on Friday, that risk materialized. Under cover of another cancellation error, the same failure brought down the net again.
The third outage occurred the same afternoon, and the validator was restarted to install the second outage fix. Upon reboot, the nodes started without any knowledge that the distributed key generation protocol (the mechanism that initializes the randomness of the network at the start of each node) was working. epoch) failed because its data was not persistently saved.
The queue of transactions that depend on that randomness grows unresolved, and the closing logic epoch (a network validation period that requires clearing its queue before closing) was stuck waiting for a process that never completed.
This failure affected the price of the SUI token. It was trading at $0.95 on May 28, when the issue began. The stock fell to a low of $0.90 and closed the day at $0.92. The note was trading at US$0.87 at close. This is 8.4% below the level before the decline began.
As reported by CriptoNoticias, the waterfall Sui experienced between May 28 and 29 was not the first incident of its kind. Interruptions were recorded in November 2024 and January 2026.
Structural problems revealed by the incident
A post-mortem examination identified two structural issues that were revealed by the fall. has to do with the first one Accumulated complexity of modules that calculate and charge fees for each transaction. According to the SuTeam report, the code is currently “so complex that borderline cases (situations that the code was not explicitly designed to handle) are difficult to filter out through code review alone,” making it difficult to predict failures before they occur.
Second, when a defective transaction (one that the validator cannot process without generating an irrecoverable error) enters the system, the network has no mechanism to contain the damage. Stop all validators at the same time.
Vadim, an on-chain developer and researcher, directly points out that second problem. According to their analysis, the trigger for the initial crash was the address balance feature introduced in version 1.72. This means that the gas fees for transactions that were canceled due to insufficient funds were also debited, resulting in a negative balance (a state that the network could not resolve), which caused all validators to abort at once. “Bugs have never been a problem. What’s missing is a firewall,” he wrote in X.
The report also mentions four network improvements: epoch shutdown resiliency, gas-filled module quality, and fault isolation and diagnostic tools. However, the foundation did not provide a timeline for its implementation..
Finally, the Sui Foundation said that “user funds were not at risk” during the outage, and the network did not reverse confirmed transactions upon resumption.
(Tag translation) Altcoin
