StakeThat +MEV went into one of Solana’s first Alpenglow epochs with about 137,000 SOL staked with it and 0.8169 SOL in its vote account. The new admission ticket required 0.8198 SOL. It was 0.003 SOL short.
That’s the first post-activation picture of Solana’s Validator Admission Ticket. Most stake made it through without trouble, but a small group of validators keeps dropping in and out, often by amounts smaller than a typical transaction fee. At least one large stake program changed its delegation code hours before activation so its stake wouldn’t stay with those validators. Marinade’s published validator snapshots for epochs 1063 to 1067 show who fell short and what happened next.
The First Admission Tickets Came Due
According to Solana’s BLS Pubkey and VAT page, Alpenglow has been active on mainnet since epoch 1064, which began around 13:36 UTC on October 21. From then on, each validator’s vote account pays a 0.8 SOL Validator Admission Ticket at the start of every epoch. The fee is burned, and it pays for the following epoch. The first charge, at the start of epoch 1064, covered epoch 1065.
The rules are strict. If the fee can’t be paid, Solana’s page says, “your validator does not vote or produce blocks in the next epoch.” Validators without a registered BLS key, or outside the top 2,000 by stake, are also excluded. Delegators to an excluded validator earn no inflation rewards for that epoch. The page also stresses that the balance “has to stay funded, not be funded once.”
The vote account also has to stay above its rent-exempt minimum. In Marinade’s data, the full requirement works out to 0.8197612 SOL.
How Many Validators Fell Short
Marinade publishes the inputs to its stake auction every epoch in a public pipeline repository. The input data checks whether each validator’s vote account holds enough to cover the next epoch’s ticket. Each snapshot covers about 670 validators and roughly 438 million SOL of stake.
Before activation, in the epoch 1063 snapshot, 33 validators holding 7.63 million SOL, or 1.74% of stake, were below the ticket balance. Once Alpenglow went live, that number fell quickly: 15 validators with 1.55 million SOL (0.35%) in epoch 1064, then 19 with 1.15 million (0.26%) in epoch 1065, the first epoch the ticket actually paid for. By epoch 1066 it was 13 validators with 790,000 SOL (0.18%), and by epoch 1067 just 8 with 710,000 SOL (0.16%).
The epoch 1063 count matches a figure in Marinade’s own code history. A commit on October 21 notes that its new gate “refuses the 33 validators with vat_eligible false” in the 1063 snapshot.
The biggest pre-activation gaps were fixed in time. An unnamed validator with 2.05 million SOL of stake, Temporal Emerald with 1.23 million and Bitget Wallet with 593,000 were all below the ticket balance in epoch 1063. All three had topped up by the next snapshot. Of the 33 flagged before activation, 25 had recovered by epoch 1064.
So the fee didn’t remove a large share of stake. The balance check affected less than 2% of stake at its peak, and about 0.16% by epoch 1067.
Missing by Fractions of a SOL
The validators that did fall short show a pattern: many of them nearly had enough.
- StakeThat +MEV had 0.8139 SOL in epoch 1064 and 0.8169 SOL in epoch 1065. It was short by 0.006 SOL, then by 0.003 SOL.
- CyberVillage had 0.8166 SOL in epoch 1065, which was 0.003 SOL short. It recovered in epoch 1066, then fell short again in epoch 1067.
- SteakStache followed the same pattern: short in 1065, funded in 1066, short again in 1067.
Because the ticket is charged every epoch, a validator that tops up just enough to pass once will fail again an epoch later unless it adds more. Between epochs 1064 and 1067, between 3 and 11 validators fell below the threshold each epoch, while a similar number recovered.
The shortfalls also cut across setups. Every validator flagged in epoch 1067 runs either Agave 4.3.x, the client release the migration required, or the latest Firedancer build. Software readiness wasn’t the issue. Keeping the balance funded was.
Solana offers a fix: routing block revenue to the vote account through solana vote-update-commission-collector. That keeps the account funded automatically, though the change takes two epochs to apply.
Marinade Wrote Its Own Rule Around the Fee
The bigger story is how stake programs reacted. Marinade, one of Solana’s largest delegation programs, changed its auction code in pull request #153, merged at about 06:47 UTC on October 21, roughly seven hours before Alpenglow activated.
The change adds a new disqualifying condition, vatIneligible. A validator whose vat_eligible flag is false “gets no SAM and no backstop stake,” because “Agave then pays its stakers no inflation rewards for the epoch.” A null value never disqualifies a validator. The check currently tests only the SIMD-0357 balance, not BLS key registration.
The first version applied the rule only in Alpenglow epochs. Marinade’s developers then caught a timing problem. Agave starts charging the ticket at the 1063-to-1064 boundary, and “a stake move made in epoch E takes effect in E+1,” so the original rule “left Marinade stake on refused validators for epochs 1064 and 1065.” The merged version refuses ineligible validators in every epoch.
The snapshots confirm it worked. In the epoch 1064 snapshot, 192,762 SOL of Marinade stake sat on ineligible validators, nearly all of it on one unnamed validator. By epoch 1065, that had dropped to 2,407 SOL, and by epoch 1066 to 10 SOL. A later commit gives the rule its own label and call to action in Marinade’s interface, so operators can see why they’ve been excluded.
Marinade, in short, turned a protocol fee into a delegation rule of its own. A validator that misses one ticket loses an epoch of rewards, and it also loses Marinade stake until its balance recovers.
Were the Readiness Warnings Borne Out?
Before activation, Solana Compass reported that a community tracker flagged 45% of stake as not ready two days ahead of epoch 1064, mostly validators on Agave 4.3.0 and 4.3.1. The following day, that had dropped to 20.3%.
Marinade’s snapshots show the software gap closing much as those warnings suggested it needed to. Stake on Agave 4.3.0 and 4.3.1 fell from 6.56% in epoch 1063 to 0.08% by epoch 1067. Firedancer operators, who had to switch to Agave for the migration, dropped from 11.0% of stake to 5.9% in epoch 1064. They then came back strongly, reaching 17.2% by epoch 1067, well above their pre-migration share.
The large readiness gap was about software and keys, and it was mostly closed by the time it mattered. The fee gap was always small.
What the Ticket Changes for Delegators
For most delegators, nothing changed: more than 99.8% of stake sits with validators that keep their tickets paid. For those delegated to the few that don’t, the cost is clear: no inflation rewards for that epoch, and in Marinade’s case, stake moved elsewhere.
The protocol sets one requirement: 0.8 SOL in the vote account every epoch. Stake programs are now adding their own rules on top, starting with Marinade’s. Validators that keep running near the threshold will keep losing epochs of rewards, and possibly delegations, until they keep a buffer or automate funding.
Disclosure: This is not trading or investment advice. Always do your research before buying any cryptocurrency or investing in any services. Follow us on X @nulltxnews
Be the first to comment