Solana entered vote-only mode as block finalization stalled

Solana’s Mainnet Beta cluster entered a degraded, vote-only state on February 25, 2023, stopping economic activity while core contributors investigated slow block finalization and validators prepared a coordinated restart.

The official incident ledger places the transition at 05:46:16 UTC. In vote-only mode, validators continued producing consensus votes but stopped including ordinary economic transactions. The distinction matters: the chain had not simply become slower for users. Its safety behavior had effectively suspended transaction processing while the network attempted to recover.

By 12:32 UTC, Solana’s status service said a coordinated restart was underway. The cluster did not return to normal operation before February 25 ended in UTC.

What was known on February 25

The precise cause was not established during the event. Solana’s initial notice said protocol contributors were still identifying the root cause. Contemporaneous reporting described major transaction disruption, sharply reduced throughput and instructions for validators to restart from a selected point in the ledger.

Operators initially examined the transition from validator software version 1.13 to 1.14 because the degradation followed that upgrade. That timing supported an investigation, not a verified causal conclusion. Claims circulating on February 25 that a particular release bug had definitively caused the incident went beyond the evidence then available.

The response also illustrated how Solana recovered from a cluster-wide failure. Although validators are independently operated, restoring progress required enough operators to agree on a restart point and compatible software. Coordination was therefore a practical dependency even though no single operator could restart the network alone.

Why the interruption mattered

Solana was designed around high transaction capacity and short confirmation times. A safety mode that excluded economic transactions tested a more fundamental property: whether applications could rely on the base network being available when users needed it.

For exchanges, wallets, decentralized applications and infrastructure providers, the immediate concern was operational rather than speculative. Deposits, withdrawals, trades and other on-chain instructions could not proceed normally while the cluster remained degraded. The surviving records do not provide a complete, independently verified count of failed or delayed transactions, affected users or economic losses, so this reconstruction does not estimate them.

The incident also exposed the tension between rapid validator-software deployment and network stability. Performance improvements can change behavior only after a sufficient share of stake adopts them, making a supermajority transition materially different from testing a release on a smaller set of machines. On February 25, however, it remained uncertain whether the upgrade itself, another component or an interaction among systems had produced the failure.

Status at the end of the date

The verified February 25 record supports three conclusions: finalization slowed severely; validators entered vote-only mode and stopped processing economic activity; and a coordinated restart was attempted. It does not support a definitive event-day root-cause claim.

There is also no market-price assertion in this file. Digital-asset prices vary across venues and continuous UTC trading windows, and the cited incident records do not establish that any SOL price movement was caused by the disruption.

Later technical context

This section uses information unavailable on February 25. Solana’s incident ledger says the community restarted the cluster at 01:28 UTC on February 26 and marked the incident resolved at 02:09 UTC.

A Solana Foundation report published on April 18, 2023 attributed the degradation to abnormal traffic involving an excessively large block and custom block-forwarding services. Re-forwarded data overwhelmed deduplication logic and congested Turbine, Solana’s primary block-propagation protocol. The report said the restart rolled back no finalized transactions of economic value and identified mitigations incorporated into validator versions 1.13.7 and 1.14.17. Those findings clarify the failure but should not be projected into what participants knew during the February 25 response.