/ status

Service status

Live operational view of the mirror network. Updated from internal probes every 60 s. For the public RSS feed of incidents subscribe to /feed/incidents.atom.

All systems operational. snapshot taken 2026-05-07 09:54 UTC

Component status

HTTP resolver (edge)
operational
latency p50 81 ms · p95 460 ms
rsync daemon
operational
317 active sessions · 41.2 Gbps
Object storage (cache)
operational
1.24 PB / 1.80 PB used · 68%
Upstream sync (huggingface.co)
operational
last delta 14 min ago · 2.1 TB
Metadata index (Postgres)
operational
replication lag 220 ms
Checksum verifier
degraded
backlog 4,120 jobs · catching up

Peer mirrors

Peer Location Role Uplink 90d uptime Last probe State
fra1.hf-mirror.eu Frankfurt, DE primary · sync source 100 Gbps 99.998% 7 s ago up
ams1.hf-mirror.eu Amsterdam, NL peer 40 Gbps 99.991% 9 s ago up
waw1.hf-mirror.eu Warsaw, PL peer 25 Gbps 99.974% 11 s ago up
sto1.hf-mirror.eu Stockholm, SE peer 25 Gbps 99.962% 14 s ago up

Bandwidth · last 24 h (aggregate egress)

10:0014:0018:0022:0002:0006:0009:54
peak 78.4 Gbps · 04:18 UTC avg 41.7 Gbps total 36.1 TB / 24h

Incident history (last 90 days)

When (UTC) Component Severity Duration Summary
2026-04-22 11:08 peer waw1 minor 38 m Routing flap at upstream transit; traffic shifted to fra1+ams1. post-mortem
2026-03-09 02:14 upstream sync minor 2 h 11 m huggingface.co LFS endpoint returned 503 for ≈ 2 h. No data loss; sync caught up afterwards.
2026-02-14 19:42 edge resolver fra1 major 17 m nginx config rollout regression; 502 on ~6% of requests. Reverted. post-mortem