The runbook's WAF test now uses a Node fetch one-liner (connection pool):
xargs curl only reached ~20 requests/s, below the rule's 300 per 10 s.
Measured 2026-10-27: 800 requests in 4.9 s -> 111 x 200, 201 x Worker 429,
488 x WAF 1015.
[skip-discuss]
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Measured on production after the 4.5.0 deploy: with a new connection per
request, 1,000 requests in 53 s got no 429 from RL_SITE (120/60 s); over
one keep-alive connection, 429 after ~93. The ratelimits binding counts per
machine of a data centre and syncs late, so the zone's WAF rule is the
limit that holds against floods. Verified once the owner set the rule
(300/10 s): 800 requests in 4.9 s gave 111 x 200, 201 x Worker 429 and
488 x WAF 1015, with the site answering again 10 s later.
Runbook test commands, code comments, CLAUDE.md gotcha 43, the 4.5.0
changelog entry.
[skip-discuss]
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every request that runs the Worker is counted per client (IPv4, or IPv6
/64): RL_SITE 120/min (app pages, public /api, the old apps' /data JSON,
misses), RL_IMG 600/min (covers), RL_ADMIN 300/min (/admin + /api/admin);
RL_SUGGEST 5 and RL_GATE 20 stay on top for their POSTs. Verified bots are
not counted. Over a limit: 429 + Retry-After 60 (JSON for the API, covers,
files and /data with CORS; a script-less EN/VI page for app pages). Fails
open.
Every rate-limit namespace moves to 8801-8806 (staging 8901-8906):
namespace ids are account-wide, and other Workers on the account use
1001-1006, so RL_SUGGEST / RL_GATE shared their counters.
docs/operations.md section 1 step 15: the zone's WAF rate-limiting rule as
the first layer (Free plan: one rule, IP, 10 s; 300 requests / 10 s).
[skip-discuss]
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>