The System

arcane@prod:~$ systemctl status --all

Live telemetry from the machine that is serving you this page, right now. Nothing here is mocked or cached for show — these are the operator’s own dashboards, simply made public.

A quick map of what each panel tells you:

metrics — how busy the platform is at this moment: requests per second, latency, error rate, memory
history · 30m — the same numbers as sparklines, so you can read the last half hour at a glance
deploy feed — real events from the CI pipeline; a push to main lights it up, stage by stage
logs — a live, redacted tail of the platform’s own logs
trace your request — the actual network path your request just took: edge → tunnel → proxy → backend
topology — every pod and its state, straight from the Kubernetes API
deep scry — heavier diagnostics that print the exact queries they run, so you can verify them yourself

metrics

source: prometheus · live

req/s0.16

p99 latency7.8 ms

error rate0.00%

web cpu0.001 cores

web mem71 MiB

deploy feed

Merge fix/seam-breathingreached build, released
Merge fix/mobile-elevator-rework: verified elevator geometry + fluidityreached build, released
Merge fix/clearance-unclampedreached build, released
Merge fix/elevator-clearancereached build, released
Merge fix/tall-card-hintsreached build, released
Merge fix/contact-emailreached build, released
Merge fix/hero-identityreached build, released
Merge fix/pdf-cachekeyreached build, released
Merge fix/ux-batch: mobile elevator + topo + full-width prose + PDF framingreached build, released
Merge fix/lead-paragraphs: leads split into readable paragraphsreached build, released
Merge fix/bio-widthreached build, released
Merge fix/pdf-preview: iframe + frame-src self + full-width bioreached build, released

trace your request

scrying your route…

history · 30m

source: prometheus range · live

req/s0.16
web cpu0.001 cores
web mem71 MiB

logs

source: scrying…

reading the platform’s logs…

deep scry — the queries are the exhibit

cpu by pod (cores)

no data

sum by (pod) (rate(container_cpu_usage_seconds_total{namespace=~"gipc|data|observability",container!=""}[2m]))

memory by pod (MiB)

no data

sum by (pod) (container_memory_working_set_bytes{namespace=~"gipc|data|observability",container!=""}) / 1048576

request rate by status code (req/s)

no data

sum by (code) (rate(caddy_http_request_duration_seconds_count{handler="subroute"}[2m]))

latency ladder (ms)

no data

p50: histogram_quantile(0.5, sum by (le) (rate(caddy_http_request_duration_seconds_bucket{handler="subroute"}[5m]))) * 1000
p95: histogram_quantile(0.95, sum by (le) (rate(caddy_http_request_duration_seconds_bucket{handler="subroute"}[5m]))) * 1000
p99: histogram_quantile(0.99, sum by (le) (rate(caddy_http_request_duration_seconds_bucket{handler="subroute"}[5m]))) * 1000

5xx ratio (caddy) (%)

no data

5xx: (sum(rate(caddy_http_request_duration_seconds_count{handler="subroute",code=~"5.."}[5m])) or vector(0)) / sum(rate(caddy_http_request_duration_seconds_count{handler="subroute"}[5m])) * 100

node root-fs used (%)

no data

used: (1 - sum(node_filesystem_avail_bytes{mountpoint="/"}) / sum(node_filesystem_size_bytes{mountpoint="/"})) * 100

node network (physical NICs) (KiB/s)

no data

rx: sum(rate(node_network_receive_bytes_total{device!~"lo|veth.+|cni.+|flannel.+|tailscale.+|cilium.+|lxc.+"}[2m])) / 1024
tx: sum(rate(node_network_transmit_bytes_total{device!~"lo|veth.+|cni.+|flannel.+|tailscale.+|cilium.+|lxc.+"}[2m])) / 1024

log volume by app (lines / 5m)

grafana3.0

kube-state-metrics30.0

ollama33.0

prometheus1.0

sum by (container) (count_over_time({namespace=~"gipc|data|observability"} [5m]))