Solana remained operational through an infrastructure failure that temporarily disrupted part of its validator network on Aug. 12, according to Solana Foundation technology executive Jacob Creech.
Summary
- 597 of 699 staked Solana validators kept voting while the network continued producing blocks normally.
- 102 validators temporarily stopped voting, while affected operators recovered within forty minutes after the disruption.
- 28.83% of staked SOL went delinquent, approaching the 33.34% threshold where transaction finality stops entirely.
- Teraswitch said a malformed route disrupted twelve European and Asian sites before engineers restored connectivity.
- Solana’s status page recorded no mainnet incident and shows 100% cluster uptime across ninety days.
Of 699 staked Solana validators, 597 continued voting while blocks and transactions continued to be processed. Affected validators recovered within 40 minutes.
Creech said validators in the Solana Foundation Delegation Program were unaffected. Solana’s official status page recorded no mainnet incident on Aug. 12 or Aug. 13 and showed 100% Mainnet Beta cluster uptime over the previous 90 days.
Solana validators kept finality despite a sharp drop
While the mainnet stayed online, separate analysis showed the incident came closer to disrupting finality than the raw validator count suggests. Marinade Finance found 28.83% of all staked SOL became delinquent for about 33 minutes. Solana requires more than two thirds of stake to participate for transactions to reach finality, putting the relevant offline threshold at 33.34%.
Marinade identified roughly 90 validators affected by the routing failure, while Creech’s 597 of 699 figure means 102 validators were not voting at one point. The difference reflects separate measurements rather than evidence that all 102 validators shared the same infrastructure failure.
Creech described the incident as a “proof point for Solana’s resiliency.” The network did withstand the disruption, but Marinade’s data also showed the delinquent stake reached about 86% of the level at which finality would have stopped.
Teraswitch routing fault spread from Miami to Asia
Teraswitch’s status report traced the infrastructure problem to a malformed default route originating from its MIA1 facility in Miami. A route reflector in Amsterdam propagated the altered route into European and Asia Pacific markets, where local routers preferred it over valid routes. Twelve sites in London, Amsterdam, Dublin, Frankfurt, Singapore and Tokyo lost reachability. North American sites were not affected.
Engineers identified the malformed route within 10 minutes and removed Miami from the private backbone. Service returned at 04:16:15 UTC. Teraswitch later deployed a global change across its compute sites so a similarly malformed route should not block traffic forwarding. The provider said the underlying defect remains under investigation and a full root cause report will follow.
The event also exposed infrastructure concentration among validators. Marinade calculated that one autonomous system held about 118.9 million SOL, or more than one quarter of all staked SOL, and roughly 94% of that stake went offline together.
Solana avoids a repeat of its 2024 network halt
The outcome contrasts with the network’s February 2024 restart after block production stopped. Validators required a coordinated restart during that incident, and Solana remained offline for nearly five hours. The Aug. 12 infrastructure failure did not require a mainnet restart.
Solana has since pursued additional resilience through independent validator software. In related coverage, Firedancer began producing Solana mainnet blocks in 2026, adding another validator client path alongside the dominant Agave ecosystem.
The latest disruption tested a different type of decentralization: physical hosting and network connectivity rather than validator software. Solana continued processing transactions, but the concentration of stake behind common infrastructure allowed one provider failure to remove a large share of voting stake at once.
What happens next
Teraswitch’s immediate configuration fix is already deployed, but its investigation is not finished. The provider is still examining why the Miami default route was advertised with incorrect attributes and has engaged its hardware vendor. A full report is expected once that work is completed.
Validator operators are also likely to face closer scrutiny over infrastructure redundancy. Marinade said it plans to review concentration limits by autonomous system and data center and provide more transparency around automatic failover arrangements. For Solana, the next test is whether those infrastructure changes reduce the share of stake exposed to any single routing failure.
Be the first to comment