P50 lamports——
P75 lamports——
P95 lamports——
P99 lamports——
slot—mainnet-beta
samples0ring buffer
staleness—vs feed stamp
sol/usd—jupiter lite
awaiting first poll
measured

Feed status

Every bar below is a probe this browser made, not a stored uptime record. There is no server-side history yet, and inventing one would be a lie.

Live probes

/api/tips · GETcache 15s

Jito tip_floor percentiles in lamports

probing…1 bar = 5s probe, this browser session, newest right
/api/health · GETcache none

current slot + upstream latency

probing…1 bar = 5s probe, this browser session, newest right
/api/rpc · POST getEpochInfocache 30s

whitelisted Solana RPC passthrough

probing…1 bar = 5s probe, this browser session, newest right

Known behaviour

surfaceobservedwhat we do
getRecentPrioritizationFeespublic RPC answers 0 for every fee levelnot used as a data source; the wall reads Jito tip_floor instead
/api/tipsserver cache 15s, upstream refresh slower than thatthe status bar shows seconds since the feed timestamp, not since our fetch
/api/price without ids400 bad idsdocumented in the API reference, kept as-is
/api/rpc403 for any method outside the whitelistintentional, smallest possible surface
SOL/USDJupiter lite v3USD columns blank out rather than guess when the price is missing

Incident log

No incidents recorded. This log is hand-maintained, starts empty, and is not an uptime claim. Timestamps will be UTC when there is something to write.

SLOs

Targets, not achievements: percentile feed younger than 30s; proxy p95 under 1.5s; sample ring intact across reloads. Misses get written here, including the boring ones.