Routing failure caused nearly 30% of Solana staked assets to go offline—dangerously close to stopping the network
On August 12, 2026, the Solana network experienced a large-scale, temporary validator outage due to a routing configuration error by the managed service provider Teraswitch. The incident briefly pushed the system toward the protocol’s configured finality interruption threshold. While the event itself did not bring the entire network down, it exposed the concentration risk of staked nodes at the network and autonomous system (AS) layer, prompting the market to revisit discussions about the resilience of high-performance blockchain infrastructure.
As for the core facts: a default route issued from Teraswitch’s Miami site lost routing metrics and community attributes during propagation. It was then disseminated to multiple nodes across Europe and the Asia-Pacific region via an Amsterdam routing reflector. Local edge routers mistakenly treated it as a local high-priority route, while the core network determined it to be invalid—leading roughly a dozen sites (including London, Amsterdam, Dublin, Frankfurt, Singapore, and Tokyo) to be unable to forward valid traffic for a time. North America was not affected. The fault caused about 28.83% of already-staked SOL to go offline briefly. The gap to Solana’s block-finality halt threshold—when offline staked assets reach approximately 33.34%—was only about 4.51 percentage points. Based on related estimates, it effectively covered about 86% of the distance toward the “network shutdown” threshold. About 90 validators were affected, with total estimated losses of roughly 333 SOL in block rewards, expected to be covered by validator margin after the epoch ends. Teraswitch’s engineering team identified the issue in about 10 minutes, and service was restored around 04:16:15 UTC.
In terms of logic breakdown, the impact was highly concentrated in a single autonomous system, AS20326. This AS carried about 118.89 million SOL—around 27.34% of the network’s total staked amount—already exceeding the 25% safety cap that the Solana Foundation set in its delegated program for a single AS. Moreover, about 94% of the nodes in this AS went offline within the same time window. In addition to Teraswitch, platforms such as latitude.sh, Limestone, Butterfly Research, and Allnodes also saw a combined total of about 14.10 million SOL go offline synchronously, suggesting that if dispersion is counted only under the data-provider/managed-provider perspective, it may underestimate the scale of simultaneous disconnections caused by shared-origin routes or regional convergence failures. Some nodes recovered in concentrated bursts during windows in Amsterdam, Frankfurt, Tokyo, and similar locations—more like waiting for routes to re-converge rather than switching to backup paths promptly. Solana’s second-largest validator, Helius, was offline for about 33 minutes as well, further indicating that the distribution of hot-standby coverage and automatic failover capability is uneven.
Pathways of impact on the crypto market: first comes the infrastructure narrative. High-performance chains emphasize throughput and low latency, but if key validators share upstream network providers or autonomous systems, a localized routing error can quickly escalate into a network-wide finality risk. Second is the staking and delegation structure: if institutions and liquid-staking protocols are overly concentrated in a small number of ASes or data centers, operational risk can be transmitted into protocol-layer availability risk. Third is trust and governance expectations: even if this incident only resulted in limited reward loss and validator margin provided a backstop, once the system crosses a one-third offline threshold, transaction confirmations would stall—and there is no equivalent margin mechanism guaranteeing coverage for systemic consequences. Historically, the last time Solana fully shut down was in February 2024; the restart took nearly 5 hours. The market’s sensitivity to how it prices “temporary disruption” versus “material interruption” is therefore not the same.
Editor’s judgment and observations: this incident looks more like a stress test near the threshold rather than a realized network disaster. In terms of facts: routing has been restored, reward losses were limited, and finality was not truly interrupted. In terms of speculation: if validators continue lacking verifiable hot switching and multi-path egress, the probability of recurrence from similar failures is unlikely to be ignored. Platforms such as Marinade have stated they will review limits on network and data-center concentration and push for disclosures about hot-standby and automatic failover information—this is more constructive than merely debating short-term price volatility. For ecosystem participants, what matters more is the real dispersion of staking geography and network topology, not the nominal number of validators. Going forward, if more protocols publish AS distributions, data-center redundancy, and results from failure drills, it will help transform the “almost a shutdown” episode into measurable governance improvements rather than repeating risk warnings.
#Solana质押因路由错误接近停机 #BTC #ETH #BNB
On August 12, 2026, the Solana network experienced a large-scale, temporary validator outage due to a routing configuration error by the managed service provider Teraswitch. The incident briefly pushed the system toward the protocol’s configured finality interruption threshold. While the event itself did not bring the entire network down, it exposed the concentration risk of staked nodes at the network and autonomous system (AS) layer, prompting the market to revisit discussions about the resilience of high-performance blockchain infrastructure.
As for the core facts: a default route issued from Teraswitch’s Miami site lost routing metrics and community attributes during propagation. It was then disseminated to multiple nodes across Europe and the Asia-Pacific region via an Amsterdam routing reflector. Local edge routers mistakenly treated it as a local high-priority route, while the core network determined it to be invalid—leading roughly a dozen sites (including London, Amsterdam, Dublin, Frankfurt, Singapore, and Tokyo) to be unable to forward valid traffic for a time. North America was not affected. The fault caused about 28.83% of already-staked SOL to go offline briefly. The gap to Solana’s block-finality halt threshold—when offline staked assets reach approximately 33.34%—was only about 4.51 percentage points. Based on related estimates, it effectively covered about 86% of the distance toward the “network shutdown” threshold. About 90 validators were affected, with total estimated losses of roughly 333 SOL in block rewards, expected to be covered by validator margin after the epoch ends. Teraswitch’s engineering team identified the issue in about 10 minutes, and service was restored around 04:16:15 UTC.
In terms of logic breakdown, the impact was highly concentrated in a single autonomous system, AS20326. This AS carried about 118.89 million SOL—around 27.34% of the network’s total staked amount—already exceeding the 25% safety cap that the Solana Foundation set in its delegated program for a single AS. Moreover, about 94% of the nodes in this AS went offline within the same time window. In addition to Teraswitch, platforms such as latitude.sh, Limestone, Butterfly Research, and Allnodes also saw a combined total of about 14.10 million SOL go offline synchronously, suggesting that if dispersion is counted only under the data-provider/managed-provider perspective, it may underestimate the scale of simultaneous disconnections caused by shared-origin routes or regional convergence failures. Some nodes recovered in concentrated bursts during windows in Amsterdam, Frankfurt, Tokyo, and similar locations—more like waiting for routes to re-converge rather than switching to backup paths promptly. Solana’s second-largest validator, Helius, was offline for about 33 minutes as well, further indicating that the distribution of hot-standby coverage and automatic failover capability is uneven.
Pathways of impact on the crypto market: first comes the infrastructure narrative. High-performance chains emphasize throughput and low latency, but if key validators share upstream network providers or autonomous systems, a localized routing error can quickly escalate into a network-wide finality risk. Second is the staking and delegation structure: if institutions and liquid-staking protocols are overly concentrated in a small number of ASes or data centers, operational risk can be transmitted into protocol-layer availability risk. Third is trust and governance expectations: even if this incident only resulted in limited reward loss and validator margin provided a backstop, once the system crosses a one-third offline threshold, transaction confirmations would stall—and there is no equivalent margin mechanism guaranteeing coverage for systemic consequences. Historically, the last time Solana fully shut down was in February 2024; the restart took nearly 5 hours. The market’s sensitivity to how it prices “temporary disruption” versus “material interruption” is therefore not the same.
Editor’s judgment and observations: this incident looks more like a stress test near the threshold rather than a realized network disaster. In terms of facts: routing has been restored, reward losses were limited, and finality was not truly interrupted. In terms of speculation: if validators continue lacking verifiable hot switching and multi-path egress, the probability of recurrence from similar failures is unlikely to be ignored. Platforms such as Marinade have stated they will review limits on network and data-center concentration and push for disclosures about hot-standby and automatic failover information—this is more constructive than merely debating short-term price volatility. For ecosystem participants, what matters more is the real dispersion of staking geography and network topology, not the nominal number of validators. Going forward, if more protocols publish AS distributions, data-center redundancy, and results from failure drills, it will help transform the “almost a shutdown” episode into measurable governance improvements rather than repeating risk warnings.
#Solana质押因路由错误接近停机 #BTC #ETH #BNB