A Routing Bug Took Solana 86% of the Way to Losing Finality
A misconfigured route at hosting provider Teraswitch briefly took 28.83% of staked SOL offline, bringing Solana within 4.51 percentage points of its finality threshold and exposing deeper concentration risks across validators and infrastructure providers.
Solana came within 4.51 percentage points of losing the ability to finalize transactions on Wednesday morning after a routing fault at hosting provider Teraswitch knocked 28.83% of staked SOL offline. The network stops finalizing blocks when roughly 33.34% of stake is delinquent, meaning the incident reached about 86% of the way to a full finality failure.
How the Routing Failure Spread
Teraswitch said a default route from its Miami site was propagated without the metric and communities that normally help its network distinguish the preferred path. A route reflector in Amsterdam then distributed the route into Europe and Asia-Pacific.
Edge routers interpreted the announcement as locally originated and preferred it over their legitimate routes. The data center core rejected the route as invalid, leaving 12 sites in London, Amsterdam, Dublin, Frankfurt, Singapore and Tokyo without a valid forwarding path. North American sites were unaffected.
Engineers identified the problem within 10 minutes, and service was restored at 04:16:15 UTC. The disruption lasted about 33 minutes. According to staking data analyzed afterward by Marinade Finance, roughly 90 validators were affected and collectively missed 333 SOL in rewards. Those losses will be covered by validator bonds at the end of the epoch.
Stake Concentration Intensified the Risk
The outage was concentrated in a single autonomous system, AS20326, which carried 118.89 million SOL, or 27.34% of total network stake. About 94% of that stake went offline simultaneously—well above the 25% ceiling set by the Solana Foundation’s delegation program for any one autonomous system.
Marinade’s analysis found that 59 validators holding 80.2 million SOL returned in a narrow window across Amsterdam, Frankfurt and Tokyo after waiting for routes to reconverge. Helius, Solana’s second-largest validator, remained down for the full incident. Marinade also identified another 14.1 million SOL that went offline across latitude.sh, Limestone, Butterfly Research and Allnodes, suggesting that hosting-provider labels alone may underestimate correlated failures.
The findings included several notable data points:
- 59 validators holding 80.2 million SOL recovered together.
- 74 operators were measurable in Marinade’s review, with three returning cleanly: Laine, Cogent Crypto and Lion3d.
- Marinade said four autonomous systems account for two-thirds of the stake distributed through its allocation model; AS395201 alone represents 36.94%.
What This Means
The incident did not halt Solana, but it demonstrated how a failure outside the blockchain’s software can threaten network finality when stake is concentrated across a small number of infrastructure systems. Had delinquent stake exceeded one-third, no transactions would have finalized for any SOL holder, and validator bonds would not have compensated that systemic loss.
Marinade said it will review concentration limits by network and data center and begin publishing whether validators use hot swap and automatic failover. The event therefore puts infrastructure diversity, not just validator count, at the center of Solana’s resilience debate—especially given that the last outright halt, in February 2024, took nearly five hours to restart.