docs introduction
The problem and incident history
Why a single sequencer halt matters, and the Base halts on record.
Base orders and executes transactions through a single sequencer. When that sequencer stops producing blocks, the chain halts: no transaction on Base is executed until block production resumes. Markets outside Base do not stop with it.
This page describes what a halt means for anyone with an open position on Base, and lists the halts referenced throughout these docs.
A single sequencer
On Base, one sequencer receives transactions, orders them and produces blocks. With a single sequencer, a halt stops block production for the whole chain at once. Transactions submitted during the halt wait; none of them is executed until the sequencer restarts.
What stops during a halt
While Base is halted, anyone with a position on Base cannot:
- Swap out of an asset that is falling, or into one they need.
- Repay a loan to reduce their debt.
- Add collateral to a position that is moving towards liquidation.
- Withdraw funds to move them somewhere else.
No fee level changes this. There are no blocks to include a transaction in until the chain restarts.
Prices keep moving elsewhere
Centralised exchanges and other chains keep trading during a Base halt. The assets that users hold, borrow or post as collateral on Base keep repricing on those venues.
When Base restarts, the first blocks meet prices that moved while users could not act. Price feeds update, and positions are marked against the moved prices. A loan that was safe when the chain stopped can be eligible for liquidation in the first blocks after the restart, and its owner competes for those blocks with everyone else trying to act at the same moment. The longer the halt, the further prices can move before anyone on Base can respond.
Incident history
The halts below are the ones referenced throughout these docs, with durations as reported on status.base.org (opens in a new tab).
| Date | Reported duration | Tiers crossed, by reported duration |
|---|---|---|
| 5 Sep 2023 | 43 min | 15 min |
| 5 Aug 2025 | 33 min | 15 min |
| 25 Jun 2026 | about 2 h | 15 min and 1 h |
| 26 Jun 2026 | under 1 h | Not 1 h or 3 h; 15 min depends on the exact duration |
- 43 min15 mincrossed1 hnot crossed3 hnot crossed
- 33 min15 mincrossed1 hnot crossed3 hnot crossed
- about 2 h15 mincrossed1 hcrossed3 hnot crossed
- under 1 h15 min?depends on the exact duration1 hnot crossed3 hnot crossed
The tier column is illustrative. STALL was not live during any of these incidents and paid nothing. The column shows which trigger tiers a halt of the reported duration would have crossed. None of the halts listed exceeded the 3 h threshold by reported duration.
A real payout would depend on the halt duration measured by the Sequencer Uptime Feed, which can differ from the duration reported on a status page. For the 26 Jun 2026 halt, reported as under 1 h, whether the 15 min tier would have been triggered depends on the exact duration the feed measured. See measuring halt duration.
Why a parametric trigger
Losses during a halt are hard to measure and harder to prove. They depend on each user's positions, on prices elsewhere, and on what each user would have done if they could have acted. A claims process would need someone to assess every one of them.
STALL avoids that by paying on an observable event instead. If the Sequencer Uptime Feed shows a halt that exceeds the threshold of a trigger tier, every active cover position in that tier pays its full payout size. The trade-off is basis risk: the payout is fixed, and it can be larger or smaller than any individual loss. How cover works explains the mechanics.