Status
Everything, including the bad days.
Component health, regional latency and every incident we have had. Updated from the probes that page our on-call. Last update 2026-09-16 05:58 UTC.
Arc Mainnet RPC
chain 5042, HTTP and WSS
Arc Testnet RPC
chain 5042002, HTTP and WSS
Archive pools
unpruned, all core regions
Unified USDC transfer feed
REST and WSS
Settlement Proof API
certificate cache and verifier
Webhooks
delivery pipeline
Streams
backfill and realtime sinks
Dashboard and Admin API
control plane
A region counts as down when p99 exceeds 2s or the error rate exceeds 1% for 60 consecutive seconds. Bars cover 90 days.
Latency
Seven day percentiles, by region.
| Region | p50 | p95 | p99 | 90 day uptime |
|---|---|---|---|---|
| us-east-1 | 11ms | 24ms | 41ms | 100.00% |
| us-west-2 | 14ms | 29ms | 48ms | 100.00% |
| eu-central-1 | 12ms | 26ms | 44ms | 99.99% |
| eu-west-1 | 13ms | 27ms | 46ms | 100.00% |
| ap-southeast-1 | 15ms | 31ms | 52ms | 99.99% |
| ap-northeast-1 | 16ms | 33ms | 55ms | 100.00% |
Incidents
Every one, with the duration and the fix.
Including the ones no customer noticed. An incident log that only contains outages people complained about is a marketing document.
Elevated p99 in ap-southeast-1
2026-09-04 · resolved · 22 minAn upgrade to arc-node v0.8.0 in Singapore ran snapshot reconciliation during peak hours and p99 rose to 310ms. No requests failed and no data was lost. Upgrades now run behind a drained pool.
Webhook delivery delay
2026-08-19 · resolved · 9 minA queue partition rebalance delayed delivery by up to 90 seconds. Every event was delivered, ordering held, nothing was dropped. Partition count raised and rebalances moved to a maintenance window.
Testnet archive lag after the Zero5 backfill
2026-07-28 · resolved · 1h 14mRe-indexing testnet history across the Zero5 emitter change put the testnet archive pool up to 400 blocks behind tip. Mainnet was unaffected. Backfills now run on a shadow pool and swap in when complete.
Subscribe by email, webhook or RSS from the dashboard. Enterprise customers are paged in parallel with our on-call.