The Sui Foundation has completed a post-mortem on three mainnet outages that hit the network on May 28 and 29, 2024. The episodes, the foundation said, stemmed from software bugs — not external attacks — and involved a gas-charging flaw in an upgrade and a risky patch that backfired.
The gas-charging culprit
The first two halts traced back to the same cause: a bug in the gas-charging logic introduced with the v1.72 'address balances' upgrade. That upgrade went live shortly before the incidents. Under certain conditions the charging routine went haywire, freezing transactions and bringing the chain to a stop. The foundation’s engineers restored service twice before identifying the exact code path that triggered the failure.
They issued a fix for the gas bug. But the trouble wasn't over.
A patch that made things worse
The third outage happened when the team applied a known-risk patch designed to address a separate randomness-state fault. That fault — a state inconsistency in the random number generator used by smart contracts — had been flagged internally as a potential issue. The patch itself, however, introduced its own failure mode. When it was deployed to the live network, the randomness state logic collapsed, taking mainnet down for a third time.
The foundation’s engineers had to roll back the patch and rebuild the fix from scratch. That work took several hours, during which the network was unreachable.
What the post-mortem revealed
The post-mortem document, released last week, lays out the timeline and root causes in technical detail. It does not name any individual or external contractor as responsible. Instead it points to gaps in the upgrade testing process, especially for changes that touch core fee and randomness logic. The foundation said it has since added more simulation scenarios and mandatory audit steps for any patch marked “high risk.”
The outages did not result in lost user funds, the foundation confirmed. But they did halt block production on a network that processes millions of dollars in decentralized finance transactions daily. Some validators reported slashing penalties during the downtime, though the foundation said it would work to reverse those.
The unresolved question
The Sui network has recovered and is processing blocks normally. But the post-mortem leaves one question open: will the revised testing protocol catch the next edge case before it hits production? The foundation says it’s now running longer dry-runs on testnet for any patch that touches consensus-critical components. The next high-risk upgrade will test whether those measures are enough.
