fix: revert required srt frontend variables

[cquil11]

Summary

• Revert the benchmarks/multi_node/agentic_srt.sh changes from config(dsv4): add GB300 Dynamo-TRT AgentX points / 添加 GB300 Dynamo-TRT AgentX 配置点 #2690.
• Stop requiring SRT_FRONTEND_HOST and SRT_FRONTEND_PORT for all existing multi-node AgentX recipes.
• Restore the frontend drain metrics probe to http://localhost:$PORT/metrics.

This addresses the compatibility regression identified in #2690 (comment).

Validation

• bash -n benchmarks/multi_node/agentic_srt.sh
• git diff --check
• Confirmed the file matches its state immediately before config(dsv4): add GB300 Dynamo-TRT AgentX points / 添加 GB300 Dynamo-TRT AgentX 配置点 #2690.

[github-actions]

Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase As a PR reviewer and CODEOWNER, I have reviewed this and have.

For PR verification, add the full-sweep-fail-fast label (strongly recommended) to this PR — the benchmark sweep only runs on labeled PRs. Use full-sweep-enabled only if you need matrix jobs to keep running past a failure.

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. See GitHub's docs on re-running failed jobs

---

感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 As a PR reviewer and CODEOWNER, I have reviewed this and have。

如需进行 PR 验证，请为此 PR 添加 full-sweep-fail-fast 标签（强烈推荐）— 基准测试 sweep 仅在带有标签的 PR 上运行。仅当需要矩阵任务在失败后继续运行时才使用 full-sweep-enabled。

PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。参见 GitHub 关于重新运行失败任务的文档

[claude]

Claude finished @cquil11's task in 2m 23s —— View job

---

Review of PR #2706

• Gather context (PR diff, referenced PR config(dsv4): add GB300 Dynamo-TRT AgentX points / 添加 GB300 Dynamo-TRT AgentX 配置点 #2690)
• Verify the revert matches the pre-config(dsv4): add GB300 Dynamo-TRT AgentX points / 添加 GB300 Dynamo-TRT AgentX 配置点 #2690 state
• Review for bugs / regressions
• Post review summary

Verification

The revert is exact: git diff 6816ed05f~1 c0c915672 -- benchmarks/multi_node/agentic_srt.sh is empty, so the file matches its state immediately before #2690, as claimed. Existing multi-node AgentX recipes (GB200/GB300 vLLM, sglang) no longer fail check_env_vars on the unset SRT_FRONTEND_HOST/SRT_FRONTEND_PORT, and their aiperf client and drain probe are back on localhost:$PORT, which works for them because they co-locate the client with the frontend. Note that #2690's companion change in benchmarks/benchmark_lib.sh:1999 (--url ${AIPERF_SERVER_URL:-http://localhost:$PORT}) was intentionally left in place; since nothing exports AIPERF_SERVER_URL anymore, it falls back to the old behavior everywhere, so it's harmless.

One follow-up to be aware of

🟡 WARNING: This revert re-breaks the two multi-decode GB300 Dynamo-TRT AgentX recipes added in #2690.

Why it matters: All six recipes under benchmarks/multi_node/srt-slurm-recipes/trtllm/deepseek-v4/agentx-gb300-20260811/ set client_placement: last_decode with orchestrator_placement: first_decode. For the Np1d topologies (1p1d, 2p1d, 3p1d, 5p1d) there is a single decode group, so the client is co-located with the frontend and http://localhost:$PORT still resolves. But for 1p4d-dep4-tep8-c4-b1-mtp and 1p6d-dep4-tep4-c24-b4-mtp, last_decode ≠ first_decode: the aiperf client and the drain metrics probe will now target localhost on a node where no frontend is listening, so those two configs will fail to connect when they next run. The SRT_FRONTEND_HOST/SRT_FRONTEND_PORT export that this PR removed was what made those placements work.

Fix: In a follow-up, restore frontend addressing in a backward-compatible way instead of making it required, e.g. in agentic_srt.sh:

if [ -n "${SRT_FRONTEND_HOST:-}" ] && [ -n "${SRT_FRONTEND_PORT:-}" ]; then
    export AIPERF_SERVER_URL="http://${SRT_FRONTEND_HOST}:${SRT_FRONTEND_PORT}"
fi

and derive the drain probe URL from ${AIPERF_SERVER_URL:-http://localhost:$PORT} — or change those two recipes to client_placement: first_decode. Fix this →

Summary

The revert itself is correct, minimal, and restores compatibility for all pre-existing AgentX recipes — no blocking issues in the change as written. The one caveat is that the #2690 1p4d and 1p6d GB300 Dynamo-TRT recipes depended on the removed export and will need the backward-compatible follow-up above (or a placement change) before their next sweep.

[claude]

This revert unconditionally restores the localhost-only frontend URL, but doing so re-breaks the 6 GB300 dsv4 dynamo-trt AgentX recipes added in #2690 that use split first_decode/last_decode placement (see inline comment for details). A conditional approach (use SRT_FRONTEND_HOST/PORT when present, fall back to localhost otherwise) would satisfy both recipe families instead of trading one regression for another.

Extended reasoning...

Overview

Single-file, small diff to benchmarks/multi_node/agentic_srt.sh reverting the SRT_FRONTEND_HOST/SRT_FRONTEND_PORT requirement and AIPERF_SERVER_URL export introduced in #2690, restoring the old localhost-only frontend assumption for the drain probe and (via benchmark_lib.sh's fallback) the replay traffic URL.

Security risks

None — this is benchmark tooling, no auth/crypto/data-exposure surface.

Level of scrutiny

This is a revert intended to fix a compatibility regression for colocated recipes, but I verified against the actual recipe YAMLs that it introduces a new regression for the 6 GB300 dsv4 split-placement recipes from the same #2690 commit (confirmed orchestrator_placement: first_decode / client_placement: last_decode with decode_nodes >= 4 in all six files, plus a corroborating comment in runners/launch_gb300-nv.sh that only makes sense if loopback doesn't resolve to the frontend in the normal AgentX path). This warrants human review before merging — the fix as written trades one broken recipe family for another.

Other factors

The inline bug comment lays out the full reproduction and a concrete suggested fix (conditional AIPERF_SERVER_URL), so a human reviewer has what they need to resolve this without re-deriving it.

[claude]

🔴 This revert hardcodes the AgentX frontend at http://localhost:$PORT (both the drain probe here and the --url fallback in benchmark_lib.sh:1999), but the 6 GB300 DeepSeek-V4 dynamo-trt recipes added in the same #2690 commit (agentx-gb300-20260811/*.yaml) set orchestrator_placement: first_decode / client_placement: last_decode with decode_nodes >= 4, so the frontend and benchmark client always run on different nodes and never set SRT_FRONTEND_HOST/PORT themselves. Reverting unconditionally re-breaks exactly the points #2690 added (traffic and drain probe both hit a non-listening localhost); a better fix is to make AIPERF_SERVER_URL conditional — set it only when SRT_FRONTEND_HOST/SRT_FRONTEND_PORT are present, falling back to localhost otherwise — so both recipe families work.

Extended reasoning...

The bug: This PR reverts agentic_srt.sh to drop the SRT_FRONTEND_HOST/SRT_FRONTEND_PORT requirement and the AIPERF_SERVER_URL export, restoring the old assumption that the AgentX frontend and the benchmark client are colocated on the same node (hardcoding http://localhost:$PORT/metrics for the drain probe, and relying on benchmark_lib.sh:1999's --url ${AIPERF_SERVER_URL:-http://localhost:$PORT} fallback for the actual replay traffic). That assumption is true for the older colocated recipes this PR is trying to fix, but it is false for the 6 GB300 DeepSeek-V4 dynamo-trt AgentX recipes under benchmarks/multi_node/srt-slurm-recipes/trtllm/deepseek-v4/agentx-gb300-20260811/, which were added in the very same #2690 commit that introduced the SRT_FRONTEND_* machinery this PR now reverts.\n\nThe code path: Verified directly against the recipe files — all six recipes set frontend.orchestrator_placement: first_decode and benchmark.client_placement: last_decode, and every one has decode_nodes >= 4 (values are 8, 8, 6, 8, 4, 4 across the six files). Since first_decode and last_decode index into a decode-node list of size >= 4, they always resolve to two different physical nodes. None of these recipes set SRT_FRONTEND_HOST/SRT_FRONTEND_PORT in their benchmark.env (only PORT: 8000 etc.) — that injection was expected to come from the external srt-slurm harness based on the placement fields, which is exactly what #2690's check_env_vars ... SRT_FRONTEND_HOST SRT_FRONTEND_PORT + export AIPERF_SERVER_URL was there to consume.\n\nWhy nothing else prevents this: build_replay_cmd in benchmark_lib.sh:1999 still reads --url ${AIPERF_SERVER_URL:-http://localhost:$PORT}; this PR only touches agentic_srt.sh, so once AIPERF_SERVER_URL is no longer exported, that fallback silently kicks in. There's also a smoking gun in runners/launch_gb300-nv.sh:252-260, which for the eval-only path explicitly rewrites orchestrator_placement: first_decode -> head with the comment 'Keep AgentX frontends on first_decode for throughput, but co-locate the eval-only frontend with lm-eval so loopback resolves.' That comment only makes sense if loopback does not resolve to the frontend in the normal (non-eval-only) AgentX path — confirming the frontend and client are genuinely on separate nodes for these recipes.\n\nStep-by-step proof (using dynamo-disagg-gb300-1p1d-dep8-dep32-c388-b4-mtp.yaml, decode_nodes=8):\n1. The srt-slurm harness allocates 8 decode nodes; first_decode = decode node #1 (runs the dynamo frontend/orchestrator on port 8000), last_decode = decode node #8 (runs agentic_srt.sh as the benchmark client).\n2. Before this PR: SRT_FRONTEND_HOST/SRT_FRONTEND_PORT are injected pointing at decode node #1; agentic_srt.sh exports AIPERF_SERVER_URL=http://<decode-node-1-ip>:8000, and both the replay traffic and the drain probe correctly target that remote node.\n3. After this PR: SRT_FRONTEND_HOST/SRT_FRONTEND_PORT are no longer required or consumed, so AIPERF_SERVER_URL is never set.\n4. build_replay_cmd falls back to --url http://localhost:8000 — but the script is running on decode node #8, where no dynamo frontend is listening on port 8000.\n5. The drain probe likewise hits http://localhost:8000/metrics on decode node #8, which also isn't serving that endpoint.\n6. Result: every request in the AgentX replay fails to connect, and/or the drain probe never gets valid metrics — the recipe fails, for all 6 of #2690's new points.\n\nThe fix: Don't revert unconditionally. Make AIPERF_SERVER_URL conditional — export it as http://${SRT_FRONTEND_HOST}:${SRT_FRONTEND_PORT} only when both vars are present (and drop them from check_env_vars's hard-required list), otherwise fall back to http://localhost:$PORT for the colocated recipes. Same for the drain probe URL. That satisfies both the older colocated recipes (the actual regression target of this PR, per the linked discussion) and the 6 split-placement GB300 dsv4 recipes from #2690, instead of trading one compatibility break for another.