7 d window · history · ends 2026-11-05 22:42Z ·
2026-10-29 22:42Z → 2026-11-05 22:42Z ·
/api/summary ·
/api/timeseries
Sparkline: line = average latency, red band = outage window
· Bucket = 35 min · Vantage: xnode-1 over public internet
Own1 host metrics
Pushed every ~60 s from Own1. Orange line = 1m load avg over 24h. Dashed = overcommit threshold (load > #cores). Red tinted zone above the threshold is when yggdrasil keepalives start to starve.
GPU is returning corrupt images — 512×512 generations come back as
noise with HTTP 200 and a valid PNG, so every other check on this
board still reads green. memegen and comicgen "square" are affected.
A reboot restores it.
events
0
reboots
853 bad gens
up 26.6d
image
BAD
512×512
sd —
15.3h ago
ygg
3/h
3 peers up
59m 6s
—
path
OK
LA ygg 20%
EU ygg 40%
net 0%
Upstream Openmesh yggdrasil relay (root-cause indicator)peer.yggdrasil.openmesh.cloud over public IPv4 — TCP handshake on TLS port 9003. If this is BAD, every yggdrasil-routed site below will also be BAD.
▸dns1.trustless.cloud NS · bridge record for ASHTON's boxnsDNS
ns
BAD
16813/30234
1199 ms
17.7h
DNS
OK
90682/90702
94 ms
—
▸23.227.167.191 US · Los Angeles — DNS :53TCP
TCP
OK
20156/20156
121 ms
—
▸46.232.249.203 EU · Nuremberg (netcup)TCP
TCP
OK
20156/20156
174 ms
—
▸92.5.225.96 NS · dns1 raw IP — no PTR, not a dedicated peerTCP
TCP
OK
10078/10078
171 ms
—
▸peer.yggdrasil.openmesh.cloud DNSTCP
DNS
OK
30205/30234
147 ms
—
TCP
OK
10058/10078
189 ms
—
Yggdrasil-routed sites (depend on the relay)CNAME → 206-…yggdrasil.trustless.cloud. Reachability depends on the relay above + the trustless.cloud DNS proxy.