# @zacodil: Base's outage shows single sequencer risk, dirty memory state bug

Source: https://firefly.social/post/x/2070939836916113533?sid=1506745023

## Summary

Base's outage, as described in a post-mortem, came from a sequencer that hit an invalid transaction while building a block and failed to clear the leftover memory state it had touched. The next valid transaction ran on that dirty state, miscalculated its gas, and produced a block whose math did not check out. Other nodes re-executed the block, detected the mismatch and rejected it, which the author calls correct behavior. The author argues that because Base has a single block producer, no other party could build a valid replacement, so the rejection itself became a two-hour outage, whereas on a chain with many producers the next block producer would simply build the next block.

## Article

Base's post-mortem confirms the uncomfortable part. 

The sequencer was building a block, hit an invalid transaction, and failed to wipe the leftover memory state it touched. The next valid transaction then ran on top of that dirty state, miscalculated its gas, and produced a block whose math didn't check out.

Every other node re-executed the block, got the correct result, saw it didn't match, and rejected it. Exactly what they're supposed to do. BUT Base has a single block producer - so all the honest nodes agreeing to reject the bad block didn't route around the fault, it was the fault. nobody else exists to build a good block.

On a chain with many producers, rejecting one bad block is a non-event, the next one just builds it. on Base it's a two-hour outage. same correct behavior, opposite outcome, because everything hangs off one machine.