Skip to content

fix(gateway): stop stale control ui auth retry loops#81771

Merged
BunsDev merged 1 commit into
mainfrom
meow/control-ui-auth-history-72139
May 14, 2026
Merged

fix(gateway): stop stale control ui auth retry loops#81771
BunsDev merged 1 commit into
mainfrom
meow/control-ui-auth-history-72139

Conversation

@BunsDev

@BunsDev BunsDev commented May 14, 2026

Copy link
Copy Markdown
Contributor

Summary

Behavior addressed

A stale Control UI token no longer keeps reconnecting every backoff interval when there is no cached device-token fallback to try. Slow/large chat history rendering is also capped by raw tool-output payload size so a dashboard tab is less able to starve channel sockets.

Real environment tested

Local Codex worktree on macOS, Node v24.13.0, pnpm v10.33.2, focused Vitest through the repo node scripts/run-vitest.mjs wrapper.

Exact steps or command run after this patch

node scripts/run-vitest.mjs ui/src/ui/gateway.node.test.ts src/gateway/client.test.ts ui/src/ui/chat/build-chat-items.test.ts
node scripts/run-vitest.mjs src/gateway/reconnect-gating.test.ts ui/src/ui/controllers/chat.test.ts ui/src/ui/views/chat.test.ts
pnpm exec oxfmt --check --threads=1 CHANGELOG.md ui/src/ui/gateway.ts ui/src/ui/gateway.node.test.ts ui/src/ui/chat/history-limits.ts ui/src/ui/chat/build-chat-items.ts ui/src/ui/chat/build-chat-items.test.ts src/gateway/client.ts src/gateway/client.test.ts
git diff --check

Evidence after fix

  • Focused gateway/chat history regression run: 4 files passed, 127 tests passed.
  • Adjacent reconnect/chat controller/view run: 4 files passed, 115 tests passed.
  • Formatter check: all matched files use the correct format.
  • Whitespace check: clean.
  • Commit signature verified locally with git log -1 --show-signature.

Observed result after fix

  • Browser Control UI client pauses after AUTH_TOKEN_MISMATCH when local storage has no trusted device token to retry, instead of creating another WebSocket.
  • Node Gateway client pauses reconnect and calls onReconnectPaused for the same no-fallback token mismatch case.
  • The trusted cached-device-token retry path remains intact.
  • Chat history windowing now includes raw tool-result content in the render-size budget and tolerates malformed history entries.

What was not tested

  • pnpm check:changed through Testbox was not completed. node scripts/crabbox-wrapper.mjs run --provider blacksmith-testbox --timing-json --shell -- "pnpm check:changed" failed because .github/workflows/crabbox-hydrate.yml does not contain a Blacksmith testbox step, then the wrapper hung and was stopped.
  • No live browser stale-token smoke was run.
  • No live Slack Socket Mode pong-timeout reproduction was run; the source-level auth loop and history-budget behavior are covered by focused tests.

@openclaw-barnacle openclaw-barnacle Bot added app: web-ui App: web-ui gateway Gateway runtime size: M maintainer Maintainer-authored PR labels May 14, 2026
@BunsDev
BunsDev force-pushed the meow/control-ui-auth-history-72139 branch from 30a276d to 934f67c Compare May 14, 2026 09:32
@clawsweeper

clawsweeper Bot commented May 14, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Codex review: needs maintainer review before merge.

Summary
The PR changes browser and Node gateway reconnect gating for AUTH_TOKEN_MISMATCH, adds bounded Control UI chat-history character-budget windowing with regression tests, and updates the changelog.

Reproducibility: yes. for source-level reproduction: current browser and Node clients can keep reconnecting on AUTH_TOKEN_MISMATCH when no stored device-token retry exists, and current chat history rendering is count-capped only. I did not reproduce the live Slack Socket Mode disconnect in this read-only pass.

Real behavior proof
Not applicable: The external-contributor proof gate does not apply to this member/maintainer-labeled PR; the body reports focused test and formatter runs but no live browser or Slack smoke.

Next step before merge
Manual maintainer review/landing is the right next action because this is a protected maintainer/member PR with no discrete automated repair finding and remaining validation choices around live/browser and broad gates.

Security
Cleared: The diff touches auth retry and UI history traversal but adds no dependency, workflow, secret, or package-resolution changes; the bounded estimator avoids the prior client-side DoS concern.

Review details

Best possible solution:

Land this or an equivalent maintainer-reviewed patch that preserves one-shot trusted device-token recovery, stops stale token-mismatch reconnect loops without a queued fallback, keeps the current 100-message cap, and closes #72139 after validation.

Do we have a high-confidence way to reproduce the issue?

Yes for source-level reproduction: current browser and Node clients can keep reconnecting on AUTH_TOKEN_MISMATCH when no stored device-token retry exists, and current chat history rendering is count-capped only. I did not reproduce the live Slack Socket Mode disconnect in this read-only pass.

Is this the best way to solve the issue?

Yes, the PR direction is the narrow maintainable fix: gate reconnects on an actually queued trusted fallback and add UI-local bounded render budgeting. The safer landing path is maintainer validation rather than automated cleanup or a separate repair PR.

What I checked:

  • protected/member PR state: GitHub context shows author association MEMBER and the protected maintainer label, so this workflow must keep the PR open for explicit maintainer handling. (934f67c2e665)
  • current browser reconnect gap: Current main only queues device-token retry when a stored token, trusted endpoint, explicit shared token, and retry hint are all present; the close handler otherwise still falls through to scheduleReconnect() for an initial AUTH_TOKEN_MISMATCH. (ui/src/ui/gateway.ts:402, 365c986a5b2a)
  • current Node reconnect gap: Current main's Node client pauses token mismatch reconnects only when the endpoint is untrusted or the retry budget has already been consumed, so a trusted endpoint with no queued stored-token retry can keep reconnecting. (src/gateway/client.ts:698, 365c986a5b2a)
  • protocol contract: Gateway protocol docs define AUTH_TOKEN_MISMATCH recovery as one bounded trusted cached-device-token retry, then clients should stop automatic reconnect loops and surface operator action guidance. Public docs: docs/gateway/protocol.md. (docs/gateway/protocol.md:723, 365c986a5b2a)
  • current chat history gap: Current main still windows chat history by the 100-message count cap only in buildChatItems, with no UI render-character budget in history-limits.ts. (ui/src/ui/chat/build-chat-items.ts:317, 365c986a5b2a)
  • PR diff addresses prior chat-budget review issues: The downloaded PR diff preserves CHAT_HISTORY_RENDER_LIMIT = 100, adds CHAT_HISTORY_RENDER_CHAR_BUDGET, counts raw content/text/args fields with WeakSet/depth/node/budget bounds, and adds tests for raw tool_result.content and malformed entries. (ui/src/ui/chat/build-chat-items.ts:327, 934f67c2e665)

Likely related people:

  • BunsDev: Prior merged history under Val Alexander touches Control UI auth failure handling and recent Control UI refresh/chat-history behavior, and this member PR is in the same surface. (role: recent area contributor; confidence: high; commits: 79ae8148f79c, 5ae385b2f0f1, 4935e24c7a7f; files: ui/src/ui/gateway.ts, ui/src/ui/chat/build-chat-items.ts, ui/src/ui/chat/history-limits.ts)
  • joshavant: Commit history shows work hardening cached-token fallback, reconnect gating, protocol docs, and tests in the affected gateway/client paths. (role: auth reconnect behavior contributor; confidence: high; commits: a76e81019333; files: ui/src/ui/gateway.ts, src/gateway/client.ts, ui/src/ui/gateway.node.test.ts)
  • vincentkoc: Recent current-main history and the related chat-budget PR context show WebChat reload/reconciliation work adjacent to the history-rendering surface. (role: recent adjacent WebChat contributor; confidence: medium; commits: cff991c88d04, 02908db62b30; files: ui/src/ui/chat/build-chat-items.ts, ui/src/ui/controllers/chat.ts)
  • steipete: Current-main blame for the affected files and commit history show recent integration/release work plus the chat item builder extraction in the same UI path. (role: recent area contributor and integrator; confidence: medium; commits: bfc798bd0b2a, 68954f9c6ccb; files: ui/src/ui/gateway.ts, src/gateway/client.ts, ui/src/ui/chat/build-chat-items.ts)

Remaining risk / open question:

  • This read-only pass did not run the PR's focused tests; correctness is based on source, diff, docs, and discussion review plus the PR body's reported commands.
  • No live browser stale-token smoke or live Slack Socket Mode pong-timeout reproduction was inspected; the source-level auth loop and UI render-budget paths are the high-confidence proof.

Codex review notes: model gpt-5.5, reasoning high; reviewed against 365c986a5b2a.

@BunsDev
BunsDev force-pushed the meow/control-ui-auth-history-72139 branch from 934f67c to d969255 Compare May 14, 2026 10:31
@BunsDev

BunsDev commented May 14, 2026

Copy link
Copy Markdown
Contributor Author

Pre-merge verification for head d9692555ee:

Local focused proof after rebasing on latest origin/main:

node scripts/run-vitest.mjs ui/src/ui/gateway.node.test.ts src/gateway/client.test.ts ui/src/ui/chat/build-chat-items.test.ts
# 4 files passed, 127 tests passed

node scripts/run-vitest.mjs src/gateway/reconnect-gating.test.ts ui/src/ui/controllers/chat.test.ts ui/src/ui/views/chat.test.ts
# 4 files passed, 115 tests passed

pnpm exec oxfmt --check --threads=1 CHANGELOG.md ui/src/ui/gateway.ts ui/src/ui/gateway.node.test.ts ui/src/ui/chat/history-limits.ts ui/src/ui/chat/build-chat-items.ts ui/src/ui/chat/build-chat-items.test.ts src/gateway/client.ts src/gateway/client.test.ts
# all matched files use the correct format

git diff --check
# clean

GitHub proof on d9692555ee:

  • Real behavior proof: run 25855231600 passed.
  • CI/check graph: run 25855232828 passed, including check, check-additional, build-artifacts, build-smoke, gateway/core/UI shards, lint, types, policy/preflight guards, and node checks selected for this diff.
  • Security High: run 25855232863 passed.
  • Critical Quality shard selection and network-runtime-boundary: run 25855232865 passed or skipped non-selected shards.
  • Path scan: run 25855232831 passed.
  • actionlint/no-tabs/generated-doc-baseline workflow: run 25855232848 passed or skipped non-applicable docs baseline.

Known proof gaps:

  • No live Slack Socket Mode pong-timeout reproduction was run.
  • No manual browser stale-token smoke was run; the browser and Node reconnect behavior is covered by focused regression tests.
  • The earlier local Testbox pnpm check:changed wrapper path was blocked because .github/workflows/crabbox-hydrate.yml lacks a Blacksmith testbox step, but the PR CI check graph for this head passed.

@BunsDev
BunsDev merged commit 348ffe6 into main May 14, 2026
112 checks passed
@BunsDev
BunsDev deleted the meow/control-ui-auth-history-72139 branch May 14, 2026 10:36
@clawsweeper clawsweeper Bot mentioned this pull request May 15, 2026
Closed
25 tasks
@clawsweeper clawsweeper Bot mentioned this pull request Jul 6, 2026
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

app: web-ui App: web-ui gateway Gateway runtime maintainer Maintainer-authored PR size: M

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

Control UI auth retry loop + blocking chat.history causes Slack Socket Mode disconnects

1 participant

@BunsDev