Solana staking payouts can still be measured, but the fee rate behind a historical reward has become harder to verify. A report published October 30 by staking provider P2P.org says the network stopped recording the commission rate applied to voting and staking rewards after September 8. The company says commissions continued to be charged and reward amounts still reconciled. The gap concerns the evidence explaining how a payout was calculated, not proof that rewards were underpaid or that validators overcharged delegators.

Why the field became empty

The change is visible in Agave, the validator client used across much of Solana. Pull request 12588, merged May 20, made the outward-facing commission field optional and set it to no value after the custom-commission-collector feature became active. The engineering rationale was that an accumulated reward can combine commission directed to different collector accounts, so a single displayed rate could be misleading. The implementation also suppresses the rate in individual stake-reward records, removing a convenient audit trail even when a holder is examining one account.

Related reporting: Lido Opens Permissionless 2,048 ETH Validator Path With 32 ETH Bond

A current rate may not explain an older payout

Historical context matters because the commission visible in a validator account today may differ from the rate saved for a prior reward period. Solana's delayed commission rules capture a rate around an epoch boundary and use that saved value during the later reward calculation. An October 18 Agave issue warns that standard vote-account views can show a validator's latest setting even when the runtime uses an earlier rate. Reading only the current value can therefore connect the wrong fee to a past reward.

Reconstruction is possible but more demanding

P2P.org says recovering the applied rate now requires replaying validator commission history and rebuilding where stake was delegated at the relevant time. Independent reporting by CryptoSlate reviewed the Agave change and found that the calculation still receives a commission input before the outward-facing record omits it. Historical tools can preserve useful observations: Jito documents validator commission history, while Marinade's snapshot parser extracts commission state and records its vintage. Those tools help reconstruct evidence, but their coverage and capture points must match the reward being audited.

What the missing record does not prove

An empty commission field does not mean a validator charged zero, and it does not establish misconduct. The open Agave issue uses a hypothetical pattern in which a validator's live rate looks low while a saved rate is high, but it does not identify a real validator following that pattern. P2P.org also sells staking and reporting services, giving it a commercial interest in the problem. Its claims are nevertheless consistent with the merged client change and the open request for better visibility.

A proposed RPC remedy remains open

The October 18 proposal asks Agave to add an optional response showing the effective inflation commission, the relevant epoch and the bank slot. That would let wallets, explorers and staking services distinguish a validator's latest setting from the rate used by the runtime. The issue was still open when checked, with no linked implementation. Until a clearer interface ships, institutional stakers and reporting teams need to preserve historical validator state, delegation evidence and reward-period context rather than relying on a current commission screen.

Why the audit trail matters

The operational consequence is narrower than a reward failure but important for funds, custodians and exchange-traded products that must substantiate staking income and fees. Net earnings can remain measurable even when the applied fee is absent, yet measuring the result is different from independently verifying the calculation. A defensible record should connect each payout to its epoch, validator delegation and effective commission evidence, while disclosing any reconstruction limits.

Sources

AI-generated editorial image; not a photograph of the reported event. Prepared with AI assistance and source verification.