Sui’s mainnet went down three times in less than 48 hours, with the latest halt striking at around 4:30 PM EDT on May 29, 2026. The third disruption was not a repeat of the original failure. It happened during deployment of the long-term patch, when validators restarted to roll out the new software and a latent bug in epoch transition handling froze user transactions again. The chain later came back online, but the sequence pushed infrastructure reliability to the center of the discussion.
The third halt was triggered during rollout of the long-term fix
The root cause of the third outage was separate from the original version 1.72 gas charging bug. According to the team’s statement on X, validators restarting to deploy the previous day’s long-term fix ran into a failure in randomness initialization, a process that starts automatically at the beginning of each epoch. That routine could not complete because it required a higher quorum threshold than the consensus system was providing at that moment.
On Sui, an epoch is a fixed window of roughly 24 hours during which the validator set remains locked. At the start of each new epoch, several initialization routines must finish before user transactions resume. Randomness initialization is one of them. Once that process failed, a latent bug in the way the failure state was preserved across validator restarts prevented the blockchain from finishing the transition into the next epoch.
The team said validators applied another fix that addressed both the underlying epoch bug and the affected epoch itself. After that, the network came back online and user transactions resumed. Mysten Labs said a detailed incident review covering all three events is still to come.
Timeline of the three outages
The three disruptions were reported between May 28 and May 30, 2026. In the first halt, SuiScan showed no new checkpoints. The cause was identified as a crash bug in gas charging logic introduced in version 1.72, and the outage lasted 5 hours and 55 minutes. The second disruption involved unresolved mainnet settlement issues. An interim fix was deployed, and the network resumed after about 1 hour, with two-thirds of validators running updated software.
The third outage arrived during an epoch change. Validators were still producing system transactions, but user transactions were not being accepted. The cause was traced to a latent bug in handling epoch transition failure states, triggered by deployment of the long-term patch. After another fix was rolled out, the network returned to normal operation.
The report also noted that Sui had previously seen a consensus divergence outage in January 2026 that lasted about six hours. This time was different: three halts unfolded across two days, with one fix leading into the next failure.
SUI drops below $1 as liquidations hit $1.88 million
The market reaction was immediate. CoinMarketCap data in the report showed SUI at $0.9035 at the time of writing, down from $0.9254 recorded after the first outage and below the $1.00 psychological level. The article put the seven-day decline at about 19%.
Derivatives traders were hit as well. CoinGlass data showed $1.88 million in long positions liquidated during the latest outage wave alone. The report added that Bitcoin remained relatively steady over the same 48-hour period, pointing to the SUI move as outage-driven rather than part of a broader crypto selloff.
One software release led to a chain of failures
All three halts were linked back to version 1.72. That release introduced address balances and free gas charging for stablecoin transactions. First came the gas charging crash. Then the long-term patch for that crash triggered the epoch failure. That failure required a third fix. The pattern described in the report is not a single isolated bug but a cascading failure sequence tied to one release.
Based on the published facts, user funds remained safe throughout the incidents. There were no forks, no lost transactions, and no rolled-back state. Validators also coordinated fixes each time. Even so, questions around deployment testing, epoch boundary handling, and the stability of randomness initialization are now harder to avoid. The incident review from Mysten Labs is expected to be the key document for assessing the broader impact of the outage wave.