前言

背景：本文彙總了 Sui 主網近期事件：發生於 5 月 28–29 日期間的三次中斷，以及 Sui 開發團隊辨識出的根本原因。目的是說明何者失效、工程師如何診斷事件，以及隨後的影響與緩解措施。 重點：分開但重疊的軟體錯誤——一組與 gas 收費邏輯和一個新的地址餘額功能相關，另一組與驗證者在重啟時保存狀態有關——結合起來在 48 小時內造成超過 15 小時的停機。下文旨在為尋求了解這些故障及其運營影響的讀者提供清晰、中立的技術敘述。

要點彙整

關鍵重點： 三個中斷時段，合計超過 15 小時；前兩次由發行版本 1.72 中新增的地址餘額功能與既有 gas 收費邏輯之間的衝突所致，導致驗證者當機。第三次中斷源自另一個影響重啟期間隨機性狀態的驗證者層級錯誤。Sui 基金會選擇快速的臨時修復以恢復服務，同時開發人員著手完善的解決方案。市場影響包括 SUI 價格在一週內出現顯著下跌。

正文

The Sui mainnet experienced three distinct outage events between May 28 and May 29, which together amounted to over 15 hours of downtime. Engineers investigated and traced the incidents to two categories of software failures. The first two outages were linked to interactions between the network's gas-charging mechanisms and a newly introduced feature in Sui release 1.72 that tracks address balances. The introduction of that feature altered internal state and calculations in ways that were not fully compatible with existing gas logic. When validators attempted to process certain transactions under the new codepath, the mismatched expectations triggered crashes and caused nodes to halt.

These failures illustrate a common risk during protocol upgrades: even minor changes to state representation or accounting can ripple into core subsystems such as gas metering. In this case, the gas-charging logic relied on assumptions that the new address-balances code violated under particular conditions. When multiple validators hit the same conditions, they diverged from peers and stopped validating, producing the first two outage windows.

The third outage, which occurred on Friday afternoon, had a different technical origin. During a scheduled epoch change, an unrelated latent bug in how validators preserve randomness state across restarts caused synchronization problems. Validators maintain internal randomness or entropy-related state that is essential for coordinated operations. If that state is not consistently preserved or restored after a restart, validators can fall out of sync, which may block epoch transitions or other coordination points. Although distinct from the gas-related crashes, this validator-level state-management bug compounded the overall reliability picture by creating another period of downtime.

The Sui Foundation acknowledged these incidents and explained the remediation approach. To bring the network back online quickly, developers implemented an interim fix that accepted a higher short-term operational risk in favor of stability. This stopgap allowed validators to return to service while engineers developed and tested a more comprehensive patch. The Foundation communicated that user funds were not at risk during the outages and that no committed transactions were reverted — an important reassurance for users and integrators. Still, the outages represent a material reliability lapse for a mainnet that launched in 2023 and has since experienced other major incidents, including a two-hour transaction scheduling bug in November 2024 and a six-hour consensus divergence in January 2026.

The market response was notable: SUI’s price declined over the week, falling approximately 18% in the seven days surrounding the incidents. That drop occurred within a broader market downturn, including weakness in major assets such as Bitcoin, but the outages and the perception of elevated protocol risk likely contributed to downward pressure on SUI.

From a technical and operational standpoint, these outages highlight several lessons for blockchain projects performing live upgrades. First, changes that affect fundamental bookkeeping — such as balance tracking — must be validated against gas, transaction processing, and other low-level subsystems to ensure no hidden coupling. Second, validator node implementations need robust state persistence and restart logic; failures in preserving critical state can produce complex, time-consuming recovery scenarios during epoch boundaries or scheduled transitions. Third, communications and mitigation plans matter: the decision to use an interim fix demonstrates the tradeoffs teams face between minimizing downtime and ensuring a fully audited resolution.

Looking ahead, the Sui developer team is expected to release a comprehensive fix addressing both the gas interaction issues and the validator state-preservation bug. Postmortem reports, broader testing of release candidates, and enhanced upgrade procedures (for example, staged rollouts and more extensive preflight checks that simulate restart scenarios and gas calculations) can help reduce the chance of similar incidents. For node operators and integrators, staying current with node software and participating in testnets or canary deployments before major upgrades can further mitigate operational risk.

In summary, the recent outages were the product of different but intersecting software defects: an incompatibility between new address-balances code and existing gas-charging logic caused crashes, and a separate validator state-preservation bug disrupted epoch transitions. While user funds were not compromised and no committed transactions were undone, the interruptions underscore the fragility that can accompany protocol change and the importance of cautious, well-tested upgrade processes.

重點一覽表