From 2d4debd12a3e1ca7b0fd84ce1a1879f734f8e20e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?St=C3=A9phane=20Tailland?= Date: Sat, 29 Aug 2026 07:57:53 +0200 Subject: [PATCH] fix(llm): double the per-client token budget to 100k/min A tool-calling agent re-sends the whole tool result on every iteration of its loop, so a single question against a tool returning a full collection burns the old budget in three turns and answers 429 in the middle of reasoning. Pagination on those tools is the real fix; this buys room until then and stays well inside the 500k global ceiling. Co-Authored-By: Claude Opus 5 --- dev/proxy-kustomization.yaml | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/dev/proxy-kustomization.yaml b/dev/proxy-kustomization.yaml index 59036f9..9dee05f 100644 --- a/dev/proxy-kustomization.yaml +++ b/dev/proxy-kustomization.yaml @@ -40,7 +40,13 @@ spec: VECTORDB_HOST: llm-dev-vectordb VECTORDB_PORT: "6379" RATE_LIMIT_GLOBAL_TOKENS_PER_MINUTE: "500000" - RATE_LIMIT_PER_CLIENT_TOKENS_PER_MINUTE: "50000" + # Doubled from 50000. A tool-calling agent re-sends the whole tool + # result on every iteration of its loop, so one question against a + # tool that returns a full collection (list-products is ~23 KB) can + # burn the old budget in three turns and answer 429 mid-reasoning. + # Pagination on those tools is the real fix; this buys room until + # then, and stays well inside the global ceiling. + RATE_LIMIT_PER_CLIENT_TOKENS_PER_MINUTE: "100000" # Looked up after consumer.yaml's llm-anonymous KongConsumer first # reconciled — config.anonymous needs the real Consumer ID, not the # username (Konnect stores it as a literal string otherwise, silently