Solana’s network came within 14% of crashing to a halt early on Wednesday, August 12, after 28.83% of the network’s staked $SOL fell out of consensus.
The delinquency that occurred on Teraswitch, and reported by the Marinade Finance staking protocol, would have ended Solana’s 30-month uptime streak. However, the network only got 86% of the way.
If the amount of unaccounted staked $SOL on the network reached the 33.34% mark, Solana would have blown a fuse, and the network would have gone offline.
Why does the Solana network crash?
Solana stops finalizing blocks when more than one-third (33.34%) of all $SOL tokens staked on the network go delinquent. Delinquency occurs when a network validator suddenly drops out of the consensus lineup.
The network came very close to the 33.34% trip line, reaching 28.83% before things started to level out.
According to Marinade, the delinquency affected 90 validators, who lost 333 $SOL in rewards while they were offline. Ironically, at $SOL’s trading price of $76.9 as of this Cryptopolitan report, the entire Solana network could have been temporarily halted because of a $25,600 anomaly.
As the crisis was averted, Solana’s official status page still maintains its clean 100% cluster uptime for the 90 days. The network’s last full halt was on February 6, 2024, which lasted for roughly five hours.
How was the fix applied?
According to Teraswitch’s status page, “Customers at LON1, AMS1, AMS2, AMS3, DUB1, DUB2, FRA2, SGP1, SGP2, TYO1, TYO2 and TYO3 experienced loss of reachability to Internet destinations, as well as to internal Teraswitch backbone destinations between affected sites. Other North American sites were not affected.”
It added that it intentionally removed MIA1 (Miami, FL) from the backbone as part of its response.
The event, which Marinade said “barely registered anywhere,” was resolved after Teraswitch engineers “identified the malformed route within 10 minutes of onset and removed MIA1 from the backbone to halt further propagation.”
Teraswitch clarified that “affected sites reconverged on their local default routes and service was restored at 04:16:15 UTC.”
Which Solana validators went down?
Of the 74 validators that Marinade measured, only three recovered cleanly: Solana Strategies’ laine and Cogent Crypto, as well as Lion3d.
The rest of the validator pool went down for some portions of the episode. Solana’s second-largest validator, Helius, was down for the entire 33-minute disruption.
As for the lost 333 $SOL rewards, Marinade said operators will absorb the loss, with stakers getting zero exposure.
Solana’s uptime streak almost went to zero
Two days before this near-miss, Solana had gone 30 straight months without a network-wide outage. The root cause then was a bug in the LoadedPrograms JIT cache that forced validators into repeated recompilation until consensus stalled on a single block, per Solana Compass. Anza patched it, and the chain resumed five hours later.
The reliability case rests on three changes shipped across 2023 and 2024: the QUIC transport layer with stake-weighted quality of service to throttle spam, a priority fee market that now drives roughly 88% of daily fee revenue, and Firedancer, the independent validator client Jump Crypto brought to mainnet in late 2025.
Running a second client means a bug in one codebase no longer has a path to halting the whole network.
