Live network instrument · mainnet-beta

Every few hundred ms
the chain strikes
a new slot.

Slotsmith hammers raw RPC readings into one readable instrument: the live slot ticker, real throughput, slot time, skipped slots, who leads next, and exactly when this epoch cools.

Current slotconnecting…
Tx / second (last min)—
Epoch progress—
Open the forge floor What is a slot?
01

The forge floor

Live readings from solana-rpc.publicnode.com. The big counter is interpolated between polls using the measured slot time, then snapped to the real value on every reading.

Connecting to RPC…

Slot ticker

processed
—
Finalized
—
Behind tip
—
Block height
—
Last reading
—

Throughput

tx / s
Transactions per second gauge
—60-min range —

Slot time

vs. classic 400 ms
—

Average over the last 60 one-minute samples.

Skipped slots

epoch to date
—

Leader slots that produced no block, from getBlockProduction.

Last 150 finalized slots

— skipped

block producedskipped slotLeaders rotate every 4 slots — ticks mark rotation boundaries.

Epoch clock

epoch —
—complete
Slots left
—
Time left (est.)
—
Ends around
—
Total tx (all time)
—

Next at the anvil

upcoming leaders
  1. Loading the leader schedule…

Each validator identity leads 4 consecutive slots (≈1.6 s). Data: getSlotLeaders.

The last hour, minute by minute

bars = tx/s · line = slot time
———
02

How the forge works

Four strikes of the hammer, all read-only, all from one public RPC endpoint.

  1. Poll

    One batched JSON-RPC request asks for the processed slot, the finalized slot and the epoch info together — so all three numbers come from the same moment.

  2. Sample

    Every minute the validator records how many slots and transactions it saw. Sixty of those samples give throughput and the true average slot time.

  3. Count

    getBlocks lists which recent slots actually hold a block. Any missing number in the range is a skipped slot — the leader was offline, late, or out-raced.

  4. Project

    Slots left in the epoch × measured slot time = an honest ETA for the epoch boundary, which shifts as the network speeds up or slows down.

03

What you can read from it

Practical reasons people keep a heartbeat monitor open.

Is it me or the chain?

If your transactions are stuck, check the tip. A steady slot time in line with the last hour and a low skip rate means the network is fine — look at your fees, RPC or blockhash instead.

Time the epoch boundary

Stake activations, deactivations and staking rewards all happen at epoch boundaries. The epoch clock tells you roughly when the next one lands.

Spot congestion early

Rising slot times and a sudden cluster of skipped slots often show up before wallets start reporting failures.

Understand commitment

Watch the gap between processed and finalized slots — usually around 30–40 slots. That gap is why "confirmed" and "finalized" differ in wallets.

Plan bots and jobs

Turn slots into wall-clock time for timeouts, blockhash expiry (150 blocks) and scheduled tasks using a measured slot time instead of a guess.

Teach the basics

A live, visual way to explain slots, leaders and epochs to newcomers — the numbers move while you talk.

04

Anatomy of a slot

A field guide to the numbers on the forge floor.

Slots are time; blocks are work

Solana divides time into slots. Each slot is a short window — Solana was designed around roughly 400 milliseconds, and the live panel above shows what the network is actually doing today — in which exactly one validator, the leader, is allowed to produce a block. A slot is a timestamp in the schedule; a block is what the leader actually built in it. When the leader is offline, too slow, or its block loses a fork race, the slot passes empty and is counted as skipped. That is why the slot number is always ahead of the block height.

Leaders rotate in groups of four

The leader schedule is computed in advance for a whole epoch, weighted by stake: a validator with more delegated stake gets proportionally more slots. Leaders are assigned four consecutive slots at a time, so a leader rotation happens every four slot-times (about 1.6 seconds at 400 ms per slot, less when slots run faster). Because the schedule is public, clients can forward transactions straight to the upcoming leaders instead of broadcasting them everywhere.

Epochs are the long clock

An epoch on mainnet is 432,000 slots. At the classic 400 ms per slot that is exactly 48 hours; at the slot time measured live above it is proportionally shorter or longer, which is why Slotsmith projects the epoch end from measured speed instead of assuming a fixed length. Epoch boundaries matter: new stake becomes active, deactivating stake cools down, inflation rewards are paid, and a new leader schedule takes effect.

Throughput, read honestly

Each performance sample counts transactions over roughly 60 seconds. Dividing by the sample period gives transactions per second. Historically the raw count included validator vote transactions, which inflated "TPS" headlines; the RPC also reports a non-vote count, and Slotsmith shows the figure the node reports for user transactions. Throughput says how busy the chain is, not how fast an individual transaction lands.

Commitment levels

Processed means a node has seen the block. Confirmed means a supermajority of stake has voted on it. Finalized means it is rooted — 31+ confirmed blocks have been built on top and it cannot be rolled back. The gap between processed and finalized slots you see above is that safety margin — typically a few dozen slots.

05

Questions from the forge

Why does the slot counter move between polls?

To keep public RPC load low, Slotsmith only asks for the slot every 5–30 seconds. In between, it advances the counter using the measured average slot time, and snaps it back to the real value on the next reading. It never runs more than 30 slots ahead of the last real reading.

Is a skipped slot a failed transaction?

No. A skipped slot means no block was produced in that window. Transactions that would have gone into it are simply picked up by the next leaders. Your transaction only fails if it runs out of blockhash validity or errors during execution.

Why is the TPS lower than the numbers I've seen in headlines?

Headline numbers often include vote transactions, which validators send to agree on blocks. The figure here is what the node reports per sample divided by the sample length — closer to real user activity.

How accurate is the epoch end time?

It is an estimate. If slot times speed up or slow down, the end shifts. Within the last few hours of an epoch it is usually accurate to a few minutes.

What is a normal skip rate?

Network-wide skip rates are typically a few percent or lower in calm conditions. Individual validators can be much higher. Short bursts of skips in the 150-slot strip are normal; a sustained high rate is worth noticing.

Do I need a wallet?

No. Slotsmith is read-only. It never asks you to connect a wallet, sign anything, or enter any keys.

Why does the data pause when I switch tabs?

Polling stops while the tab is hidden to save requests on the shared public RPC. It resumes as soon as you come back.

Limits & honesty

  • Everything comes from one public RPC node (solana-rpc.publicnode.com). If it lags or rate-limits, readings pause and a retry button appears.
  • The big slot counter is interpolated between real readings; the "Last reading" field shows the real value's age.
  • Skip rate is network-wide for the current epoch so far, not per validator. The 150-slot strip is a tiny, noisy window.
  • Leader names are raw validator identity keys — Slotsmith does not attach names or reputations to them.
  • Polling is throttled (no faster than every 5 s) and stops when the tab is hidden.
  • Nothing here predicts prices or is advice of any kind.