EXPERIMENT (do not merge): does the tail-forwarding fix stop the Windows crash?

[petebacondarwin]

Temporary diagnostic branch. Not for merge — will be closed once it has answered its question. Companion to #15001.

This branch is #14993 plus the bisect harness. Eight rounds of bisecting on Windows CI narrowed the fixtures/dev-registry crash to a single trigger:

Peer that gets killed	Assets worker	Bound by victim	Tail consumer of victim	Crashed
worker-entrypoint-with-assets	yes	yes	yes	4/5 reps
worker-entrypoint	no	yes	no	0/3 (11 cycles)
exported-handler-with-assets	yes	no	no	0/3 (14 cycles)

Being an assets worker is not sufficient, and being bound is not sufficient. The only peer whose abrupt death aborts the survivor is the one reached over the tail path through the assets RPC-proxy hop — which is exactly where #14993 stops a departed peer's rejection from escaping unhandled.

Also ruled out along the way: forced dev-server restarts (0/3), hand-written registry churn with no live peer (0/3), idle sessions at any count (0/41), port contention, start order, CPU starvation, and traffic through the binding.

Baseline to beat here is 4/5 reps over 17 kill cycles, so this run does 8 reps of the crashing arm plus both negative controls.

[changeset-bot]

🦋 Changeset detected

Latest commit: 2fee67b

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

This PR includes changesets to release 8 packages

Name	Type
miniflare	Patch
@cloudflare/deploy-helpers	Patch
@cloudflare/pages-shared	Patch
@cloudflare/remote-bindings	Patch
@cloudflare/runtime-types	Patch
@cloudflare/vite-plugin	Patch
@cloudflare/vitest-pool-workers	Patch
wrangler	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

[github-actions]

✅ All changesets look good

[pkg-pr-new]

@cloudflare/autoconfig

npm i https://pkg.pr.new/@cloudflare/autoconfig@15006

@cloudflare/build-output-utils

npm i https://pkg.pr.new/@cloudflare/build-output-utils@15006

@cloudflare/config

npm i https://pkg.pr.new/@cloudflare/config@15006

create-cloudflare

npm i https://pkg.pr.new/create-cloudflare@15006

@cloudflare/deploy-helpers

npm i https://pkg.pr.new/@cloudflare/deploy-helpers@15006

@cloudflare/kv-asset-handler

npm i https://pkg.pr.new/@cloudflare/kv-asset-handler@15006

miniflare

npm i https://pkg.pr.new/miniflare@15006

@cloudflare/pages-functions

npm i https://pkg.pr.new/@cloudflare/pages-functions@15006

@cloudflare/pages-shared

npm i https://pkg.pr.new/@cloudflare/pages-shared@15006

@cloudflare/unenv-preset

npm i https://pkg.pr.new/@cloudflare/unenv-preset@15006

@cloudflare/vite-plugin

npm i https://pkg.pr.new/@cloudflare/vite-plugin@15006

@cloudflare/vitest-pool-workers

npm i https://pkg.pr.new/@cloudflare/vitest-pool-workers@15006

@cloudflare/workers-auth

npm i https://pkg.pr.new/@cloudflare/workers-auth@15006

@cloudflare/workers-editor-shared

npm i https://pkg.pr.new/@cloudflare/workers-editor-shared@15006

@cloudflare/workers-utils

npm i https://pkg.pr.new/@cloudflare/workers-utils@15006

wrangler

npm i https://pkg.pr.new/wrangler@15006

commit: 2fee67b

[ask-bonk]

@petebacondarwin Bonk workflow was cancelled.

View workflow run · To retry, trigger Bonk again.

[petebacondarwin]

Answered: no. With #14993 applied, the crashing arm went 8/8 over 28 kill cycles (baseline without it was 4/5 over 17). No mitigation at all, so the unhandled tail rejection is not what aborts workerd.

The controls did reproduce exactly, in an independent run:

Peer killed	Assets	Bound	Tail consumer	round 8 (no fix)	this run (with fix)
worker-entrypoint-with-assets	yes	yes	yes	4/5	8/8
worker-entrypoint	no	yes	no	0/3	0/3
exported-handler-with-assets	yes	no	no	0/3	0/3

Combined: 12/13 for the one peer, 0/6 and 0/6 for the controls. The discriminator is solid; the cause is elsewhere. Closing this branch — #14993 stands on its own merits as a diagnostics fix, and the hunt continues in #15001.