--- title: "A Sequencer Bug Took Down Coinbase's Base Network Twice in Two Days" description: "Base, Coinbase's Ethereum layer-2 network, froze twice in late June — for about 116 minutes, then 20 — after a bug in the single piece of software that orders its transactions. A post-mortem traced it to faulty handling of a rejected transaction, and reopened the debate over how 'decentralized' these networks really are." category: "Crypto" category_url: https://boursel.com/category/crypto author: "Marcus Feldman" published: 2026-06-28T05:44:20.000Z updated: 2026-06-28T05:44:20.000Z canonical: https://boursel.com/article/a-sequencer-bug-took-down-coinbases-base-network-twice-in-two-days tags: ["base", "coinbase", "ethereum", "layer-2", "sequencer", "crypto"] --- # A Sequencer Bug Took Down Coinbase's Base Network Twice in Two Days Base, Coinbase's Ethereum layer-2 network, froze twice in late June — for about 116 minutes, then 20 — after a bug in the single piece of software that orders its transactions. A post-mortem traced it to faulty handling of a rejected transaction, and reopened the debate over how 'decentralized' these networks really are. Base, the Coinbase-built blockchain, stopped cold twice in two days. The network halted for roughly **116 minutes** in one incident and about **20 minutes** in another, [Cointelegraph reported](https://cointelegraph.com/news/base-post-mortem-reveals-sequencer-bug-behind-back-to-back-outages), citing the team's post-mortem. During each freeze, no new transactions could be processed — meaning every app built on Base, from exchanges to lending protocols, was effectively offline. The Base team said no user funds were at risk. ## The terms you need A few definitions make this clear: - **Ethereum** is the major blockchain on which much of crypto runs — but it is slow and expensive at busy times. - A **layer-2** (or "L2") sits on top of Ethereum to make transactions faster and cheaper: it processes them off to the side, then posts compressed summaries back to Ethereum for final settlement. **Base** is one such L2, built on the open-source "OP Stack" maintained by Optimism, and it's the **second-largest** L2 by value, holding around **$11 billion** of user assets. - A **sequencer** is the software engine at the heart of an L2: it takes in transactions, decides their order, bundles them into blocks and pushes them toward Ethereum. On Base, Coinbase runs the **only** sequencer. ## What broke The post-mortem pinned the first outage on a flaw in the sequencer's block-building logic. When a transaction failed validation — it was malformed or otherwise invalid — the sequencer was supposed to throw it out cleanly. Instead it kept "**stale journal state**": leftover data about which accounts and storage the rejected transaction had touched. That residue then corrupted every transaction after it, and the sequencer stalled, producing no further blocks until engineers patched it. The second, shorter outage wasn't the same bug again. It was a **race condition** — a timing clash between processes restarting at once — that briefly re-froze the sequencer as the network recovered. ## Why a single sequencer is the issue This is where it gets structural. Because Coinbase runs Base's only sequencer, one software bug — or one compromised server — can stop the whole network. That isn't unique to Base: most big L2s, including Arbitrum and Optimism, still rely on a single operator-run sequencer. But it sits awkwardly with crypto's decentralization ideal. The tracker [L2BEAT](https://l2beat.com/scaling/projects/base) rates networks on a maturity scale, and Base currently sits at "Stage 1" — meaning it has some safeguards but still leans on a trusted operator for core functions. These freezes were Base's latest such stumbles, following shorter sequencer-related outages in 2024 and 2025. ## What Base says it will do Base's lead, Jesse Pollak, confirmed during the incident that no funds were lost. The published post-mortem promised two fixes: **fuzz testing** — bombarding the software with random, malformed inputs to flush out hidden bugs before they reach production — and automatic "graceful recovery" so the sequencer can restart itself without operators stepping in by hand. The bigger goal, decentralizing the sequencer across multiple independent operators, remains on the to-do list rather than shipped. The argument for patience: even when Base's sequencer goes dark, the underlying Ethereum settlement layer keeps the network's records intact, and users retain the ability to withdraw funds straight to Ethereum. The argument against: a network that depends on one company's server to order transactions can't fully claim the trust guarantees of a decentralized blockchain. June's back-to-back freezes handed the skeptics fresh evidence — and Base a concrete to-do list. ## Sources - [Base post-mortem reveals sequencer bug behind back-to-back outages](https://cointelegraph.com/news/base-post-mortem-reveals-sequencer-bug-behind-back-to-back-outages) - [Base — layer-2 scaling overview](https://l2beat.com/scaling/projects/base)