Skip to content

docs(plan): G3 done; findings #91 and #92 - #92

Merged
AlpNuhoglu merged 1 commit into
mainfrom
docs/g3-result
Oct 23, 2026
Merged

AlpNuhoglu merged 1 commit into
mainfrom
docs/g3-result

Conversation

@AlpNuhoglu

Copy link
Copy Markdown
Owner

Plan only (docs/deployment-phase2-plan.md), after PR #91 was deployed and measured in production.

Plan: G3 · Refs: Bulgu #90, #91, #92

🤖 Generated with Claude Code

Records the G3 decisions (bounded method, exact status, WebSocket
handshakes as 101 outside the histogram, 5xx-only errors measured at
the gateway, fitted buckets) and the evidence. In production, through a
port-forward to the gateway's /metrics: 22 series; 5 made-up methods
opened "other" once (44); 5 more methods on random paths left it at 44,
with requests_total{method="other",...,status="404"} at 10.

The G5 query drafts now filter on the gateway (every request is counted
again upstream) and use the 5xx-only error counter. New findings: #91
(the README's 112 ms p95 includes k6's bcrypt setup requests and sits
between two modes; fixed in G7) and #92 (WebSocket sessions in the
latency histogram; fixed in PR #91).

Plan: G3
Refs: Bulgu #90, #91, #92

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@AlpNuhoglu
AlpNuhoglu merged commit c0623b6 into main Oct 23, 2026
26 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant