Otto
your autonomous SRE · read-only · asks before it acts
live · in the cluster
Show me the golden signals for checkout, and the logs behind whatever is wrong.
✓live agent · latency, throughput, errors, saturation · 5m window
✓live agent · pod readiness and restart state
✓your observability · the matching log lines for that window
Health 20 / 100. Latency, throughput and saturation are all normal. Errors is at 25% and every point of it is pod availability, not failed requests.
20
Composite health
100 minus penalties · 50 for a broken pod, counted once
2.67 pod-minutes down
Latency
182ms
p95, client side
budget 2s
Throughput
1.2k
req/min, 60s
steady
Errors
25%
5xx + pods not ready
continuous 2m40s
Saturation
31%
max of cpu, memory
no pressure
The four signalslast 5 minutes
ongoing · 2m40s-5mnow
latencythroughputerrorssaturation
Behind the errorswhy it is not 5xx
5xx share   0.0%  every request returned 200
pods not ready  1 of 4  ImagePullBackOff
checkout-6bd5c6cdbd-vphlb  0/1  Waiting
Back-off pulling image "v2.1.0-hotfix-broken"
Error: ErrImagePull · NotFound
If Errors counted 5xx alone this panel would be green all the way through the outage. The rollout is stuck, so the old pod serves every request while the new one never starts.
show the rollout historyopen the incidentpin this view
just now · Otto · live window + your observability, nothing pre-built