[core] Only write zstd to runs whose Node.js version can decode it by VaguelySerious · Pull Request #4640 · vercel/workflow

VaguelySerious · 2026-10-05T17:49:07Z

Problem

Cross-deployment writes chose whether to compress from the target run's @workflow/core version, but which codec from the writer's own runtime. A writer on Node.js 24 therefore wrote zstd into runs pinned to Node.js 20 deployments, whose node:zlib has no zstd, and those runs failed on replay. It shows up after a project moves from Node.js 20 to a newer version while long-sleeping runs are still parked on hooks.

Change

Runs record the Node.js version of the deployment that will execute them in executionContext.nodeVersion. Same-deployment start() uses the current process. Cross-deployment start() uses the target's version, now reported in the health-check probe (nodeVersion), like hookResumeInputVersion. Bun and Deno don't record it, since their node:zlib support doesn't follow process.versions.node.
HookResumeContext gains an optional nodeVersion, mirrored from the run (fallback path, world-sim).
getRunCapabilities(coreVersion, nodeVersion) only includes zstd when the Node.js version is ^22.15.0 || >=23.8.0. getCompressionMode() maps capabilities to true (any codec) / 'gzip' / false, and compress() takes that mode, so a 'gzip' target gets gzip even with WORKFLOW_COMPRESSION_CODEC=zstd.
Runs without the field (every run created before this change) get gzip from cross-deployment writers. Gzip is readable on every supported runtime. Same-deployment writes (step outputs, snapshots) are unchanged and keep using zstd.

Follow-up needed

world-vercel's stored hook resumeContext has an allowlist of fields and doesn't carry nodeVersion yet, so resumes on the fast path will use gzip even for Node.js 24 runs until the backend mirrors the field. Gzip is correct, just somewhat more CPU than zstd.

Tests

resume-hook.test.ts: gzip for a run pinned to Node.js 20, gzip for a run without the field, zstd for Node.js 24, and the resumeContext path. The gzip cases fail against main's resume-hook.ts.
start.test.ts: nodeVersion stamped for same-deployment and probed starts, and the codec chosen from the probe.
helpers.test.ts, capabilities.test.ts, compression.test.ts: probe field, version range edges (22.14/22.15/23.7/23.8), mode mapping, gzip mode overriding the env codec.

🤖 Generated with Claude Code

Cross-deployment writes (hook resumes, cross-deployment start) picked the codec from the writer's runtime, so a Node 24 producer wrote zstd to runs pinned to Node 20 deployments, which cannot decode it. Record the creating deployment's Node.js version in executionContext and the health-check probe, mirror it into HookResumeContext, and only write zstd when it is >= 22.15 (or >= 23.8). Unknown versions get gzip. Closes #4639 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

changeset-bot · 2026-10-05T17:49:15Z

🦋 Changeset detected

Latest commit: b9db7c5

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 21 packages

Name	Type
@workflow/core	Patch
@workflow/world	Patch
@workflow/builders	Patch
@workflow/cli	Patch
@workflow/next	Patch
@workflow/nitro	Patch
@workflow/vitest	Patch
@workflow/web-shared	Patch
@workflow/web	Patch
workflow	Patch
@workflow/world-testing	Patch
@workflow/errors	Patch
@workflow/world-local	Patch
@workflow/world-postgres	Patch
@workflow/world-vercel	Patch
@workflow/astro	Patch
@workflow/nest	Patch
@workflow/rollup	Patch
@workflow/sveltekit	Patch
@workflow/vite	Patch
@workflow/nuxt	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

vercel · 2026-10-05T17:49:20Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
example-nextjs-workflow-turbopack	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
example-nextjs-workflow-webpack	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
example-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-astro-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-express-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-fastify-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-hono-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-nestjs-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-nitro-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-nuxt-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-python-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-sveltekit-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-tanstack-start-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workbench-vite-workflow	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workflow-docs	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workflow-swc-playground	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workflow-tarballs	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC
workflow-web	Ready	Preview, v0	Oct 5, 2026 6:06pm UTC

github-actions · 2026-10-05T17:49:30Z

🧪 E2E Test Results

✅ All tests passed

⚠️ Flaky E2E Tests (passed on retry)

These tests failed at least once and passed on a retry. A recurring entry here is a real race worth investigating.

completes within timeout (nextjs-turbopack · vercel-ws-transport / vercel / node / ws)
fires constructor + stream callbacks in order (nextjs-turbopack · vercel-ws-transport / vercel / node / ws)
fires constructor + stream callbacks with the tool result (nextjs-webpack · vercel-prod / vercel / node / production)
health check (CLI) - workflow health command reports healthy endpoints (nextjs-webpack · local-dev / local / node / canary)
hookDisposeTestWorkflow - hook token reuse after explicit disposal while workflow still running (nuxt · local-dev / local / node / stable)
hookGetConflictWorkflow - awaiting hook.getConflict() registers hook without payload (nextjs-webpack · local-dev / local / node / stable)
passes through LanguageModelV4ToolResultOutput from tools (sveltekit · vercel-prod / vercel / node / production)
resumes in flight during the takeover are never lost: each lands in exactly one log (nitro · vercel-http-transport / vercel / node / http)

🛠 Infra Events (absorbed by the harness)

Platform anomalies the e2e harness detected and worked around (e.g. a run the queue never picked up, replaced by a fresh run). Clustered timestamps indicate a backend blip; a steady drip indicates a platform issue worth escalating.

run-pickup-stall · runs app-generated source against a registered step (tanstack-start) · at 18:13:32Z · abandoned wrun_01M46MA9D6MF5TY16H58EB1P0Y
cold-start-warmup · suite warmup (tanstack-start) · at 18:13:40Z · abandoned wrun_01M46MA9D839X1PY7Y53NDC6A6
run-pickup-stall · hookCleanupTestWorkflow - hook token reuse after workflow completion (nextjs-webpack) · at 18:19:48Z · abandoned wrun_01M46MNRMM68520PFXDZ8AAH7F
run-pickup-stall · hookGetConflictWorkflow - awaiting hook.getConflict() registers hook without payload (nextjs-webpack) · at 18:19:55Z · abandoned wrun_01M46MNZDYBNZ9GXPY75X25WD9
run-pickup-stall · customSerializationWorkflow - custom class serialization with WORKFLOW_SERIALIZE/WORKFLOW_DESERIALIZE (nextjs-webpack) · at 18:21:14Z · abandoned wrun_01M46MRCGAH4VM5YAH1E6TNC0N

E2E Test Summary

Summary

	Passed	Skipped	Total
✅ ▲ Vercel Production	3904	875	4779
✅ 💻 Local Development	4582	551	5133
✅ 📦 Local Production	4582	551	5133
✅ 🐘 Local Postgres	4582	551	5133
✅ 🪟 Windows	342	12	354
✅ 🌐 Cross-language Conformance	68	84	152
✅ dynamic-runs	0	0	0
✅ vercel-http-transport	879	183	1062
✅ vercel-multi-region	27	0	27
✅ vercel-ws-transport	595	113	708
Total	19561	2920	22481

Details by Category

✅ ▲ Vercel Production

App	Passed	Skipped
✅ astro-node	142	35
✅ astro-quickjs	142	35
✅ example-node	142	35
✅ example-quickjs	142	35
✅ express-node	142	35
✅ express-quickjs	142	35
✅ fastify-node	142	35
✅ fastify-quickjs	142	35
✅ hono-node	142	35
✅ hono-quickjs	142	35
✅ nest-node	142	35
✅ nest-quickjs	142	35
✅ nextjs-turbopack-node	169	8
✅ nextjs-turbopack-quickjs	169	8
✅ nextjs-webpack-node	169	8
✅ nextjs-webpack-quickjs	169	8
✅ nitro-node	142	35
✅ nitro-quickjs	142	35
✅ nuxt-node	142	35
✅ nuxt-quickjs	142	35
✅ python-node	66	111
✅ sveltekit-node	161	16
✅ sveltekit-quickjs	161	16
✅ tanstack-start-node	142	35
✅ tanstack-start-quickjs	142	35
✅ vite-node	142	35
✅ vite-quickjs	142	35

✅ 💻 Local Development

App	Passed	Skipped
✅ astro-stable-node	148	29
✅ astro-stable-quickjs	148	29
✅ express-stable-node	148	29
✅ express-stable-quickjs	148	29
✅ fastify-stable-node	148	29
✅ fastify-stable-quickjs	148	29
✅ hono-stable-node	148	29
✅ hono-stable-quickjs	148	29
✅ nest-stable-node	148	29
✅ nest-stable-quickjs	148	29
✅ nextjs-turbopack-canary-node	176	1
✅ nextjs-turbopack-canary-quickjs	176	1
✅ nextjs-turbopack-quickjs-snapshot	176	1
✅ nextjs-turbopack-stable-node	176	1
✅ nextjs-turbopack-stable-quickjs	176	1
✅ nextjs-webpack-canary-node	176	1
✅ nextjs-webpack-canary-quickjs	176	1
✅ nextjs-webpack-stable-node	176	1
✅ nextjs-webpack-stable-quickjs	176	1
✅ nitro-stable-node	148	29
✅ nitro-stable-quickjs	148	29
✅ nuxt-stable-node	148	29
✅ nuxt-stable-quickjs	148	29
✅ sveltekit-stable-node	167	10
✅ sveltekit-stable-quickjs	167	10
✅ tanstack-start-node	148	29
✅ tanstack-start-quickjs	148	29
✅ vite-stable-node	148	29
✅ vite-stable-quickjs	148	29

✅ 📦 Local Production

App	Passed	Skipped
✅ astro-stable-node	148	29
✅ astro-stable-quickjs	148	29
✅ express-stable-node	148	29
✅ express-stable-quickjs	148	29
✅ fastify-stable-node	148	29
✅ fastify-stable-quickjs	148	29
✅ hono-stable-node	148	29
✅ hono-stable-quickjs	148	29
✅ nest-stable-node	148	29
✅ nest-stable-quickjs	148	29
✅ nextjs-turbopack-canary-node	176	1
✅ nextjs-turbopack-canary-quickjs	176	1
✅ nextjs-turbopack-quickjs-snapshot	176	1
✅ nextjs-turbopack-stable-node	176	1
✅ nextjs-turbopack-stable-quickjs	176	1
✅ nextjs-webpack-canary-node	176	1
✅ nextjs-webpack-canary-quickjs	176	1
✅ nextjs-webpack-stable-node	176	1
✅ nextjs-webpack-stable-quickjs	176	1
✅ nitro-stable-node	148	29
✅ nitro-stable-quickjs	148	29
✅ nuxt-stable-node	148	29
✅ nuxt-stable-quickjs	148	29
✅ sveltekit-stable-node	167	10
✅ sveltekit-stable-quickjs	167	10
✅ tanstack-start-node	148	29
✅ tanstack-start-quickjs	148	29
✅ vite-stable-node	148	29
✅ vite-stable-quickjs	148	29

✅ 🐘 Local Postgres

App	Passed	Skipped
✅ astro-stable-node	148	29
✅ astro-stable-quickjs	148	29
✅ express-stable-node	148	29
✅ express-stable-quickjs	148	29
✅ fastify-stable-node	148	29
✅ fastify-stable-quickjs	148	29
✅ hono-stable-node	148	29
✅ hono-stable-quickjs	148	29
✅ nest-stable-node	148	29
✅ nest-stable-quickjs	148	29
✅ nextjs-turbopack-canary-node	176	1
✅ nextjs-turbopack-canary-quickjs	176	1
✅ nextjs-turbopack-quickjs-snapshot	176	1
✅ nextjs-turbopack-stable-node	176	1
✅ nextjs-turbopack-stable-quickjs	176	1
✅ nextjs-webpack-canary-node	176	1
✅ nextjs-webpack-canary-quickjs	176	1
✅ nextjs-webpack-stable-node	176	1
✅ nextjs-webpack-stable-quickjs	176	1
✅ nitro-stable-node	148	29
✅ nitro-stable-quickjs	148	29
✅ nuxt-stable-node	148	29
✅ nuxt-stable-quickjs	148	29
✅ sveltekit-stable-node	167	10
✅ sveltekit-stable-quickjs	167	10
✅ tanstack-start-node	148	29
✅ tanstack-start-quickjs	148	29
✅ vite-stable-node	148	29
✅ vite-stable-quickjs	148	29

✅ 🪟 Windows

App	Passed	Failed	Skipped
✅ nextjs-turbopack-node	171	0	6
✅ nextjs-turbopack-quickjs	171	0	6

✅ 🌐 Cross-language Conformance

App	Passed	Failed	Skipped
✅ python	68	0	84

✅ dynamic-runs

App	Passed	Failed	Skipped
✅ astro-vercel	0	0	0
✅ example-vercel	0	0	0
✅ express-vercel	0	0	0
✅ fastify-vercel	0	0	0
✅ hono-vercel	0	0	0
✅ nest-vercel	0	0	0
✅ nextjs-turbopack-vercel	0	0	0
✅ nextjs-webpack-vercel	0	0	0
✅ nitro-vercel	0	0	0
✅ nuxt-vercel	0	0	0
✅ sveltekit-vercel	0	0	0
✅ tanstack-start-vercel	0	0	0
✅ vite-vercel	0	0	0

✅ vercel-http-transport

App	Passed	Skipped
✅ example	142	35
✅ express	142	35
✅ hono	142	35
✅ nextjs-turbopack	169	8
✅ nitro	142	35
✅ vite	142	35

✅ vercel-multi-region

App	Passed	Failed	Skipped
✅ nextjs-turbopack	27	0	0

✅ vercel-ws-transport

App	Passed	Skipped
✅ example	142	35
✅ express	142	35
✅ nextjs-turbopack	169	8
✅ vite	142	35

📋 View full workflow run

github-actions · 2026-10-05T17:49:31Z

📊 Workflow Benchmarks

commit b9db7c5 · Mon, 05 Oct 2026 18:34:52 GMT · run logs

Backend: vercel · app: nextjs-turbopack

Metric	Scenario	Best (ms)	P75 (ms)	P90 (ms)	P99 (ms)	Samples
TTFS	step	304 (+25%) 🔻	2252 🔴 (+27%) 🔻	2455 🔴 (+27%) 🔻	2679 🔴 (+6.2%)	30
TTFS	stream	333 (+17%) 🔻	2313 🔴 (+28%) 🔻	2542 🔴 (+38%) 🔻	2712 🔴 (+28%) 🔻	30
TTFS	hook + stream	655 (-27%) 💚	2559 🔴 (+20%) 🔻	2604 🔴 (+17%) 🔻	2849 🔴 (-33%) 💚	30
Fan-out TTFS	Promise.all(100 steps)	581 (+8.6%)	985 (+17%) 🔻	2642 (+25%) 🔻	2671 (+1.8%)	10
Fan-out TTLS	Promise.all(100 steps)	1604 (+7.9%)	4466 (+76%) 🔻	6002 (+51%) 🔻	8707 (+56%) 🔻	10
STSO	1020 steps (inline)	99 (-17%) 💚	136 (-14%)	149 (-21%) 💚	231 (-35%) 💚	1019
WO	1020 steps	133236 (-16%) 💚	133236 (-16%) 💚	133236 (-16%) 💚	133236 (-16%) 💚	1
CRTT	first chunk (pooled)	78 (+9.9%)	103 (-15%)	110 (-30%) 💚	126 (-70%) 💚	28

Streams

Scenario	CRTT 1st	p75	p90	p99	CDV max	iters
paced control (100/s, 60B)	96.5 (-1%)	152 (-27%)	196 (-65%)	349 (-81%)	133 (-48%)	10
size sweep (100/s, 160B-12KB)	89.5 (-10%)	164 (-25%)	207 (-36%)	293 (-54%)	153 (-27%)	10
replay gateway-gpt-5.4-nano-2000t (1x)	100 (+6%)	196 (+11%)	1068 (+366%)	2630 (+297%)	554 (+95%)	3
replay eve-gpt-5.6-sol-2000t (1x)	93.5 (-55%)	152 (-40%)	182 (-77%)	324 (-82%)	231 (-74%)	2
replay eve-gpt-5.6-sol-2000t (2x)	89 (-26%)	170 (-30%)	211 (-41%)	717 (-17%)	295 (-45%)	3

📈 STSO distribution vs main (inline / queue-hop histograms)

1020 steps (inline)

Cumulative STSO time: main 157838ms → this run 132922ms (Δ -24916ms, -16%)

 50-100 ms  ┃                         main   0  this   1    +1
100-150 ms  ██████████████████░░░░░┃  main 687  this 917  +230
150-200 ms  █┃█████                   main 252  this  80  -172
200-250 ms  ┃                         main  42  this  11   -31
250-300 ms  ┃                         main  16  this   1   -15
300-350 ms  ┃                         main  11  this   5    -6
350-400 ms  ┃                         main   5  this   2    -3
400-450 ms  ┃                         main   1  this   2    +1
450-500 ms  ┃                         main   1  this   0    -1
500-550 ms  ┃                         main   1  this   0    -1
550-600 ms  ┃                         main   1  this   0    -1
600-650 ms  ┃                         main   1  this   0    -1
850-900 ms  ┃                         main   1  this   0    -1

📈 CRTT drill-down vs main (RTT distributions & profiles)

variant  RTT 1ms→5s+             avg         p50           p90           p99     n
control  ······▃█▁····  130.9 (-32%)  127 (-12%)    196 (-65%)    349 (-81%)  3000
sweep    ······▂█▁····  136.1 (-17%)  128 (-11%)    207 (-36%)    293 (-54%)  3000
gw 1x    ·····▁▄█▁▁▁▁·    201 (+30%)  117 (-16%)  1068 (+366%)  2630 (+297%)  5295
eve 1x   ·····▁▃█▁▁···  130.7 (-44%)  120 (-18%)    182 (-77%)    324 (-82%)  5186
eve 2x   ······▂█▂▁···  145.6 (-26%)  136 (-16%)    211 (-41%)    717 (-17%)  7779

RTT over stream progress (avg per tenth of stream, bars scaled min→max):

control  ▄▂▃▄▅▄█▅▃▁  119–147ms
sweep    ▅█▆▅▃▆▃▄▇▁  120–150ms
gw 1x    ▂▁▁▁▂█▆▁▁▁  119–514ms
eve 1x   ▅▁▃█▂▇▇▇▅▇  124–135ms
eve 2x   ▅▂▄▇▁▄▅█▆▄  130–161ms

RTT by chunk size (avg per log size bin, ~160B → ~12KB serialized, bars scaled min→max):

sweep  ▇█▇▅▅▄▁  134–137ms

Delivery jitter over stream progress (avg positive CDV per tenth of stream, bars scaled min→max):

control  ▄▁▄▂▇▆█▂▃▃  37–45ms
sweep    ▂▅▄▄▄█▃▅▆▁  52–69ms
gw 1x    ▃▂▂▁▆█▂▁▄▂  37–60ms
eve 1x   ▅█▂▆▅▃▄▁▅▅  24–30ms
eve 2x   ▇█▆▄▇▅▄▁▂█  21–30ms

ℹ️ Metric definitions & methodology

_{Streams: first-chunk RTT (the stream-open path, before any buffering/backpressure), CRTT percentiles, and worst delivery stall (CDV max). Cells are medians across iterations; per-run values in the artifacts. No 🔴/🟢 marks until targets attach.}

_{The collapsed STSO distribution section above buckets every step gap, split inline (same warm process — pure framework overhead) vs queue-hop (fresh process — dispatch, reinit, replay). █ = main, ┃ = this run, ░ = fill.}

_{The collapsed CRTT drill-down: per-variant RTT histograms (fixed log bins, · = empty) and mean RTT/positive-CDV profile lines over stream progress and chunk size. Histograms, avgs, and profiles merge exactly across runs; p50–p99 are percentile-of-percentiles. Per-index rows live in the artifacts.}

_{Best/P75/P90/P99 deltas compare against the most recent benchmark run on main at the time of this run. 🔻 flags a delta worse than +15%, 💚 one better than −15%.}

_{Metrics — TTFS: time to first step body (in-deployment start() → first step body) · Fan-out TTFS: fan-out time to first step (in-deployment start() → first of the parallel step bodies to complete) · Fan-out TTLS: fan-out time to last step (in-deployment start() → last of the parallel step bodies to complete, i.e. when the Promise.all resolves) · STSO: step-to-step overhead (gap between consecutive step bodies) · WO: workflow overhead (whole-run time outside step bodies, in-deployment anchored) · CRTT: chunk round-trip time (per-chunk write → read latency, one clock domain: deployment → stream backend → same deployment) · CDV: chunk delay variation / delivery jitter (inter-arrival gap minus inter-write gap per seq-adjacent pair; skew-free; the row is each run's MAX positive value, so one stall moves it)}

_{Scenarios — step: one trivial no-op step, no stream; no hooks, so the run stays in turbo mode (in-process fast path) · stream: one streaming step; no hooks, so the run stays in turbo mode (in-process fast path) · hook + stream: registers a hook before one step, which exits turbo mode (dispatch path) · 1020 steps: 1020 trivial sequential steps; STSO is measured between consecutive steps in the given step ranges, and WO is the whole-run overhead outside step bodies · Promise.all(100 steps): 100 trivial no-op steps started together in a single Promise.all; Fan-out TTFS is the first of them to complete and Fan-out TTLS the last, both from the in-deployment clientStart, so their gap is the spread the runtime adds across the fan-out · paced control (100/s, 60B): the control: 300 tiny (~60B) deltas metronome-paced at 100/s — zero workload structure, so it reads the transport floor and flush cadence, and disambiguates transport-wide vs workload-specific when a replay row moves · size sweep (100/s, 160B-12KB): same pacing as the control with deltas padded in rotation across seven log-spaced sizes (~160B–12KB) — rotation decouples size from stream position, so it isolates whether chunk size causes latency · replay gateway-gpt-5.4-nano-2000t (1x): raw provider SSE cadence captured at the AI gateway boundary (gpt-5.4-nano, the most popular gateway model; per-token deltas p50 208B = the modal production chunk size), replayed exactly as measured — the typical customer's workload; its CDV is the typical customer's real delivery jitter · replay eve-gpt-5.6-sol-2000t (1x): a captured eve turn (gpt-5.6-sol, the most-used demanding eve model; ~2000 output tokens = production p50 turn length) replayed exactly as measured — eve's envelope protocol re-ships the cumulative message so sizes ramp 142B→13KB; the demanding outlier tenant's reality · replay eve-gpt-5.6-sol-2000t (2x): the same eve capture at 2x — the headroom/stress row; real fast-tier models emit the same chunk sizes at proportionally higher rate, so time compression is a faithful speed model · first chunk (pooled): every run's seq-0 RTT pooled across all stream scenarios — the first chunk precedes any workload differentiation, so pooling samples one shared stream-open path with exact percentiles}

_{Replay cadences (semantic sha256) — eve-gpt-5.6-sol-2000t eaf22f5946e7c61f3c65c7006d550df180cfabd4e706254a09f22aec0cfb420d · gateway-gpt-5.4-nano-2000t 6f24ac518b6b83ff1d0e85a5fe78230db192716d66a7fc6b2fe022752001d041}

_{🔴 marks a percentile over its target (within target is left unmarked). Targets (p75/p90/p99, ms) — TTFS 200/300/600}

_{All timestamps are deployment-side; runs are triggered in-deployment, so the CI runner and api.vercel.com sit outside every measured window. TTFS = start() → first step body (includes dispatch + any cold start); Fan-out TTFS/TTLS = first/last step completion of one Promise.all from the same anchor (the gap is the runtime’s fan-out spread); STSO/WO between step bodies; CRTT inside the workflow (excludes the api.vercel.com read path).}

_{Cold starts stay in the numbers (real bursty-workload latency, inflates P75+); Best is the warm floor.}

github-actions · 2026-10-05T17:53:17Z

Sim World

Simulated world deterministic testing for races. Traces

🟠 world-sim scenario book — 1 fail of 42 total

fence=per-spec

scenario	outcome	events	virt	replay
✅ `smoke-no-steps`	completed	3	0ms	ok
✅ `smoke-one-step`	completed	6	0ms	ok
✅ `hook-at-step-started`	completed	12	0ms	ok
✅ `hook-at-step-completed`	completed	12	0ms	ok
✅ `hook-at-hook-created`	completed	12	0ms	ok
✅ `deadline-hook-wins`	completed	7	1.0h	ok
✅ `deadline-expires`	completed	7	1.0h	ok
✅ `step-vs-timer-early-settlement`	completed	8	1.0h	ok
✅ `long-sleep`	completed	11	30.0d	ok
✅ `hook-never-arrives`	stalled	3	0ms	skipped
✅ `step-retries-twice`	completed	10	2.0s	ok
✅ `parallel-steps`	completed	9	0ms	ok
✅ `hook-on-execution-state`	completed	12	0ms	ok
✅ `peek-hook-before-branch`	completed	12	0ms	ok
✅ `peek-hook-after-branch`	completed	12	0ms	ok
✅ `peek-hook-at-registration`	completed	12	0ms	ok
✅ `race-hook-before-probe`	completed	12	0ms	ok
✅ `race-hook-after-probe`	completed	12	0ms	ok
✅ `race-duplicate-delivery`	completed	13	0ms	ok
✅ `attr-hook-before-step`	completed	11	0ms	ok
✅ `attr-hook-after-step`	completed	11	0ms	ok
✅ `attr-from-step-body`	completed	13	0ms	ok
✅ `fork-hook-after-timeout`	completed	14	1.0m	ok
✅ `fork-hook-before-timeout`	completed	14	1.0m	ok
✅ `count-hook-after-timeout`	completed	17	1.0m	ok
✅ `count-hook-before-timeout`	completed	20	1.0m	ok
✅ `stale-read-step-count-fork`	completed	20	1.0m	ok
✅ `stale-read-equal-step-counts`	completed	14	1.0m	ok
✅ `step-vs-step-fork`	completed	12	0ms	ok
✅ `step-vs-step-fork-fenced`	completed	12	0ms	ok
✅ `fence-catches-benign-direction`	completed	12	5ms	ok
✅ `in-flight-before-decision`	completed	17	1.0m	ok
❌ `in-flight-before-decision-counted`	completed	17	1.0m	ok
✅ `in-flight-after-decision`	completed	19	2.0m	ok
✅ `stale-read-step-count-fork-fenced`	completed	20	1.0m	ok
✅ `fork-hook-wins`	completed	13	1.0m	ok
✅ `fork-timeout-wins`	completed	13	1.0m	ok
✅ `unclaimed-payload-under-fork`	completed	17	1.0m	ok
✅ `claimed-payload-under-fork`	completed	17	1.0m	ok
✅ `writers-independent-step-bodies`	completed	12	0ms	ok
✅ `writers-scripted-tempo`	completed	12	0ms	ok
✅ `cancel-mid-step`	cancelled	7	0ms	skipped

Full trace: world-sim.txt

github-actions · 2026-10-05T17:53:45Z

Framework	Flow route	Step reg.	Framework output
hono	267.3 KiB (+105 B)	97.4 KiB (+78 B)	2.01 MiB (+762 B)
nextjs-turbopack	274.7 KiB (+100 B)	426 B (±0)	986.5 KiB (+143 B)

About these numbers

Sizes are gzip; parentheses show the change against main.
Flow route and Step reg. gate this job, on raw bytes rather than the gzip shown, at max(2%, 50.0 KiB). Framework output is informational.

b9db7c5 · run

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

TooTallNate

AI review: approving, nothing blocking.

The diagnosis is right. The decision to compress was gated on the target, but the codec was picked from the writer's runtime. Splitting that into a CompressionMode (true / 'gzip' / false) fixes it with a small change, and making 'gzip' take precedence over WORKFLOW_COMPRESSION_CODEC=zstd is the correct way round. When the Node.js version is unknown it fails safe to gzip, which every supported runtime can read. That covers legacy runs, Bun/Deno, edge-like runtimes without process.versions, probe timeouts, and backends that don't mirror the field yet.

What I checked:

ZSTD_NODE_RANGE (^22.15.0 || >=23.8.0) matches when zlib.zstd* landed (23.8.0, backported to 22.15.0).
The only cross-deployment payload writers that consult target capabilities are start() and resumeHook(). step-executor and suspension-handler write from the run's own deployment, so they correctly keep true.
executionContext is z.record(z.string(), z.any()) in @workflow/world, so worlds won't strip the new nodeVersion key. HookResumeContextSchema gains the optional field, and the changeset covers @workflow/world. world-sim is private, so it needs no changeset.
Locally: the touched core test files pass (319 tests), and tsc --noEmit on core is clean after rebuilding @workflow/world.

Non-blocking notes:

Same-deployment resumeHook() calls on the world-vercel fast path also drop to gzip until the backend mirrors nodeVersion into the stored resumeContext, even though the writer and the run share a runtime. It's correct, just extra CPU. If the follow-up looks like it'll take a while, consider a short-circuit: when resumeContext.deploymentId equals the current deployment, use true.
The same-deployment start() stamps the caller's process.versions.node. That's sound on Vercel, where the Node.js version is per-deployment, but it assumes the caller and the flow handler share a runtime. A short comment saying so would help the next reader.
Tiny nit: getCurrentNodeVersion() has an empty try/catch around a property read. It's harmless, but unless a throwing process getter has actually been seen somewhere, it could go.

Signed-off-by: Peter Wielander <mittgfu@gmail.com>

github-actions · 2026-10-05T18:08:11Z

No backport to stable for e71d1c7 (AI decision).

This is a genuine correctness fix, but the functionality it fixes does not exist on stable: payload compression is a 5.x feature (SerializationFormat.GZIP/ZSTD, gated at 5.0.0-beta.18). Verified that packages/core/src/serialization/compression.ts and packages/core/src/serialization/codec.ts are absent on origin/stable, and that stable's capabilities.ts and resume-hook.ts contain no zstd/gzip/compression code at all, so no stable run can be written a zstd payload. The commit also depends on main-only APIs (CompressionMode, getCompressionMode) and touches docs/content/docs/v5/ and packages/world-sim/, neither of which exists on stable.

To override, re-run the Backport to stable workflow manually via workflow_dispatch and paste this commit SHA into the ref input:

e71d1c7671308ae20371c80bd769ad3a9ce123fa

Catches the docs up with SDK changes that landed on Oct 5 without (complete) docs updates. | Change | Source PR | Docs edit | | --- | --- | --- | | zstd is only written to runs whose deployment's Node.js can decode it | #4640 | The encryption page said "gzip, with zstd support in the format". It now states when zstd vs. gzip is used and links `WORKFLOW_COMPRESSION_CODEC` | | Aborting a public writable keeps the accepted prefix and doesn't close the shared stream | #4605 | Added to the existing `releaseLock()` / `close()` callout in Streaming | | Released stream writers retire their WebSocket and continue over HTTP | #4605 | Short paragraph under `WORKFLOW_STREAMS_TRANSPORT` on the Vercel World page, including the "hold the lock for a burst" tip from the `world-vercel` README | I reviewed the other Oct 5 PRs and they need no further docs: NestJS (#4604) shipped its own docs, the `@vercel/queue` bump (#4648) and `world-local` hook fix (#4636) aren't user-facing in the docs, and #4632/#4618 are docs PRs. The Workflows pages in `vercel/front` were also checked and need no changes for these. ## Docs Preview | Page | Preview | | --- | --- | | Encryption | [Compression](https://workflow-docs-git-workflow-docs-cleanup-tue-6-oct.vercel.sh/docs/how-it-works/encryption#compression) | | Streaming | [Writing to another run's stream](https://workflow-docs-git-workflow-docs-cleanup-tue-6-oct.vercel.sh/docs/foundations/streaming#writing-to-another-runs-stream) | | Vercel World | [`WORKFLOW_STREAMS_TRANSPORT`](https://workflow-docs-git-workflow-docs-cleanup-tue-6-oct.vercel.sh/worlds/vercel#workflow_streams_transport) | 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: vercel[bot] <35613825+vercel[bot]@users.noreply.github.com> Co-authored-by: Pranay Prakash <1797812+pranaygp@users.noreply.github.com>

vercel Bot deployed to Preview – workflow-docs October 5, 2026 17:52 View deployment

VaguelySerious marked this pull request as ready for review October 5, 2026 17:52

VaguelySerious requested a review from a team as a code owner October 5, 2026 17:52

[core] Allow nodeVersion in the HTTP health-check e2e assertion

f285903

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

vercel Bot deployed to Preview – workflow-docs October 5, 2026 17:59 View deployment

TooTallNate approved these changes Oct 5, 2026

View reviewed changes

VaguelySerious commented Oct 5, 2026

View reviewed changes

Comment thread .changeset/zstd-node-version-gate.md Outdated

Apply suggestion from @VaguelySerious

b9db7c5

Signed-off-by: Peter Wielander <mittgfu@gmail.com>

vercel Bot deployed to Preview – workflow-docs October 5, 2026 18:04 View deployment

VaguelySerious merged commit e71d1c7 into main Oct 5, 2026
41 of 42 checks passed

VaguelySerious deleted the peter/zstd-node-version-gate branch October 5, 2026 18:05

github-actions Bot mentioned this pull request Oct 5, 2026

Version Packages #4580

Merged

This was referenced Oct 5, 2026

Version Packages ben-vargas/workflow#104

Closed

Version Packages Thegreatsura/workflow#1

Open

pranaygp mentioned this pull request Oct 6, 2026

[docs] Catch up docs with recently shipped SDK changes #4668

Merged

Navigation Menu

Repository navigation

[core] Only write zstd to runs whose Node.js version can decode it - #4640

[core] Only write zstd to runs whose Node.js version can decode it#4640
VaguelySerious merged 3 commits into
mainfrom
peter/zstd-node-version-gate

VaguelySerious commented Oct 5, 2026

Uh oh!

changeset-bot Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

vercel Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 5, 2026 •

edited

Loading

Uh oh!

TooTallNate left a comment

Uh oh!

Uh oh!

Uh oh!

github-actions Bot commented Oct 5, 2026

Uh oh!

This branch was successfully deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

VaguelySerious commented Oct 5, 2026

Problem

Change

Follow-up needed

Tests

Uh oh!

changeset-bot Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🦋 Changeset detected

Uh oh!

vercel Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🧪 E2E Test Results

⚠️ Flaky E2E Tests (passed on retry)

🛠 Infra Events (absorbed by the harness)

E2E Test Summary

Uh oh!

github-actions Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📊 Workflow Benchmarks

Uh oh!

github-actions Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Sim World

Uh oh!

github-actions Bot commented Oct 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

TooTallNate left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

github-actions Bot commented Oct 5, 2026

Uh oh!

This branch was successfully deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

changeset-bot Bot commented Oct 5, 2026 •

edited

Loading

vercel Bot commented Oct 5, 2026 •

edited

Loading

github-actions Bot commented Oct 5, 2026 •

edited

Loading

github-actions Bot commented Oct 5, 2026 •

edited

Loading

github-actions Bot commented Oct 5, 2026 •

edited

Loading

github-actions Bot commented Oct 5, 2026 •

edited

Loading