fix(datachannel): checkpoint lobby token refresh and llm resubscribe … by Tianhui-Han · Pull Request #4818 · webex/webex-js-sdk

Tianhui-Han · 2026-03-30T07:09:54Z

COMPLETES #< INSERT LINK TO ISSUE >

This pull request addresses

https://jira-eng-gpk2.cisco.com/jira/browse/SPARK-763776

by making the following changes

Lobby admission
After being admitted from the lobby, the user must actively request and obtain the session token to connect to backend services.
Practice session promotion
Attendees in a practice session do not receive PS tokens initially; when promoted to panelist they must actively request the PS token to gain panel access.
Reconnect and resubscribe
On reconnect the client reestablishes the LLM connection and may need to resubscribe to previous channel

Change Type

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to change)
Documentation update
Tooling change
Internal code refactor

The following scenarios were tested

< ENUMERATE TESTS PERFORMED, WHETHER MANUAL OR AUTOMATED >

The GAI Coding Policy And Copyright Annotation Best Practices

I certified that

I have read and followed contributing guidelines
I discussed changes with code owners prior to submitting this pull request
I have not skipped any automated checks
All existing and new tests passed
I have updated the documentation accordingly

Make sure to have followed the contributing guidelines before submitting.

aws-amplify-us-east-2 · 2026-03-30T07:20:30Z

This pull request is automatically being deployed by Amplify Hosting (learn more).

Access this pull request here: https://pr-4818.d3m3l2kee0btzx.amplifyapp.com

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1348a535cd

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-03-31T05:45:02Z

packages/@webex/plugin-meetings/src/meeting/index.ts

    });
    this.locusInfo.on(LOCUSINFO.EVENTS.SELF_ADMITTED_GUEST, async (payload) => {
      this.stopKeepAlive();
+      await this.ensureDefaultDatachannelTokenAfterAdmit();


Do not block admitted-guest flow on token prefetch

Awaiting ensureDefaultDatachannelTokenAfterAdmit() inside the SELF_ADMITTED_GUEST listener makes MEETING_SELF_GUEST_ADMITTED emission and subsequent updateLLMConnection() depend on a network token fetch. If /datachannel/token is slow or temporarily unavailable, the client can remain in a lobby-like state longer than necessary even though locus has already admitted the user. The admission event path should not be gated by this best-effort prefetch.

Useful? React with 👍 / 👎.

We should not block admit process before get token. Maybe call it just before updateLLMConnection()

…nnect_reconnect_subscription

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 50a55c2bae

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-02T06:48:33Z

packages/@webex/plugin-meetings/src/webinar/index.ts

+    const refreshedPracticeSessionToken = await this.ensurePracticeSessionDatachannelToken(meeting);
+
+    if (refreshedPracticeSessionToken) {
+      practiceSessionDatachannelToken = refreshedPracticeSessionToken;
+    }
+
    // @ts-ignore - Fix type
    return this.webex.internal.llm


Re-check practice-session state after async token refresh

updatePSDataChannel() now awaits ensurePracticeSessionDatachannelToken() and then unconditionally calls registerAndConnect(...) using the earlier session state, but there is no second isJoinPracticeSessionDataChannel() guard after the await. If practice mode/panelist status flips while the refresh request is in flight (for example, host ends practice session), a stale invocation can reconnect the practice LLM channel after a newer invocation already cleaned it up, leaving captions/subscriptions attached to the wrong session.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6d70cb8651

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-02T08:41:07Z

packages/@webex/plugin-meetings/src/meeting/index.ts

+      this.webex.internal.llm.setDatachannelToken(
+        fetchedDatachannelToken,
+        DataChannelTokenType.Default
+      );


Revalidate meeting identity before caching fetched lobby token

ensureDefaultDatachannelTokenAfterAdmit() writes the fetched token into the shared LLM token cache after an awaited network call, but it never verifies that the meeting/locus is still the same one that initiated the fetch. If the user leaves (or switches meetings) while fetchDatachannelToken() is in flight, this stale token can overwrite the cache after teardown; the next admitted-guest flow then sees a token present and skips fetching the correct one, so updateLLMConnection() can reconnect with invalid credentials.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-02T08:41:07Z

packages/@webex/plugin-meetings/src/meeting/index.ts

  private handleLLMOnline = (): void => {
-    // @ts-ignore
-    this.webex.internal.llm.off('online', this.handleLLMOnline);
+    this.restoreLLMSubscriptionsIfNeeded();
+


Remove persistent LLM online listener on meeting cleanup

After this change, handleLLMOnline no longer unregisters itself, which makes the listener effectively persistent for the Meeting instance. Teardown still calls cleanupLLMConneciton() without removeOnlineListener, so ended/disposed meetings can keep online callbacks attached and receive future online events from later connections, causing duplicate MEETING_TRANSCRIPTION_CONNECTED triggers and stale-instance callbacks over time.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c5a1f77643

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-02T09:24:47Z

packages/@webex/plugin-meetings/src/meeting/index.ts

+      this.ensureDefaultDatachannelTokenAfterAdmit().then(() => {
+        this.updateLLMConnection();
+      });
+
      this.updateLLMConnection();


Serialize post-admit LLM reconnect attempts

This handler now invokes updateLLMConnection() twice (immediately and again in the ensureDefaultDatachannelTokenAfterAdmit().then(...) path). When prefetch resolves quickly (for example, token already cached or token-auth feature disabled), both calls can run concurrently while isConnected() is still false, causing two registerAndConnect() register requests for the same session. Because LLMChannel.register() persists whichever binding/websocket metadata returns last, this race can leave cached routing metadata out of sync with the actual connected socket and break later routed sends; these reconnect attempts should be serialized or the second call should only run when a new token was fetched.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-02T09:24:47Z

packages/@webex/plugin-meetings/src/webinar/index.ts

+      this.webex.internal.llm.setDatachannelToken(
+        datachannelToken,
+        dataChannelTokenType || DataChannelTokenType.PracticeSession
+      );


Revalidate state before caching refreshed practice token

ensurePracticeSessionDatachannelToken() stores the refreshed token in the shared LLM cache before updatePSDataChannel() performs its stale-invocation checks. If practice eligibility or meeting lifecycle changes while refreshDataChannelToken() is in flight (e.g., user leaves and cleanup clears tokens), this late write can repopulate cache with a stale token; subsequent updates then read a cached token and skip refresh, leading to practice-session register attempts with invalid credentials. Guard this cache write with a fresh state/sequence check tied to the current meeting context.

Useful? React with 👍 / 👎.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 16219deb81

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-03T01:41:31Z

packages/@webex/plugin-meetings/src/webinar/index.ts

      this._pendingOnlineListener = null;
    }

+    const refreshedPracticeSessionToken = await this.ensurePracticeSessionDatachannelToken(meeting);


Revalidate default LLM connectivity after token refresh await

In updatePSDataChannel(), default-session connectivity is checked and _pendingOnlineListener is removed before await ensurePracticeSessionDatachannelToken(meeting), but there is no second default isConnected() check afterward. If the default LLM session drops while the token refresh is in flight, this stale invocation can still run registerAndConnect(..., LLM_PRACTICE_SESSION) even though the prerequisite connection is gone, and because the pending online listener was already cleared there is no automatic retry when default comes back online.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-04-03T01:41:31Z

packages/@webex/plugin-meetings/src/meeting/index.ts

+      }
+
+      // @ts-ignore
+      this.webex.internal.voicea.updateSubchannelSubscriptions({subscribe: ['transcription']});


Await transcription resubscribe call in online handler

restoreLLMSubscriptionsIfNeeded() wraps updateSubchannelSubscriptions() in a synchronous try/catch, but updateSubchannelSubscriptions is async and its promise is not awaited or .catched here. If that promise rejects (for example, during reconnect races), the rejection bypasses this catch and becomes unhandled, so the intended warning log and graceful recovery path do not run.

Useful? React with 👍 / 👎.

…nnect_reconnect_subscription

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9d20822121

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-04-03T06:57:44Z

packages/@webex/plugin-meetings/src/webinar/index.ts

+      latestPracticeSessionDatachannelUrl !== practiceSessionDatachannelUrl
+    ) {
+      return undefined;


Reconnect with latest practice URL after token refresh

The post-refresh stale check returns when latestPracticeSessionDatachannelUrl !== practiceSessionDatachannelUrl, but this branch does not schedule a follow-up reconnect. If the practice-session URL rotates while ensurePracticeSessionDatachannelToken() is in flight (and no role/practice-status transition happens), this invocation exits and the practice LLM channel can stay disconnected until an unrelated later trigger. Instead of returning here, continue with the latest URL or immediately re-invoke updatePSDataChannel().

Useful? React with 👍 / 👎.

github-actions · 2026-04-03T07:37:42Z

🎉 Your changes are now available!
Released in: `v3.12.0-next.11`
📖 View full changelog →

Packages Updated	Version
`webex`	`3.12.0-next.11`
`@webex/plugin-meetings`	`3.12.0-next.7`

Thank you for your contribution!
🤖 This is an automated message. For questions, please refer to the documentation.

Tianhui-Han force-pushed the feat/datachannel_in_lobby_register_and_disconnect_reconnect_subscription branch from 9500d92 to d04a422 Compare March 30, 2026 07:25

fix(subchannel): cover several case

1348a53

Tianhui-Han force-pushed the feat/datachannel_in_lobby_register_and_disconnect_reconnect_subscription branch from d04a422 to 1348a53 Compare March 31, 2026 02:57

Tianhui-Han marked this pull request as ready for review March 31, 2026 05:39

Tianhui-Han requested review from a team as code owners March 31, 2026 05:39

chatgpt-codex-connector bot reviewed Mar 31, 2026

View reviewed changes

mickelr added the validated If the pull request is validated for automation. label Apr 2, 2026

mickelr requested changes Apr 2, 2026

View reviewed changes

Merge branch 'next' into feat/datachannel_in_lobby_register_and_disco…

50a55c2

…nnect_reconnect_subscription

chatgpt-codex-connector bot reviewed Apr 2, 2026

View reviewed changes

fix(subchannel): not block admit process

6d70cb8

chatgpt-codex-connector bot reviewed Apr 2, 2026

View reviewed changes

fix(subchannel): recheck join status after await in updatePSDataChannel

c5a1f77

chatgpt-codex-connector bot reviewed Apr 2, 2026

View reviewed changes

fix(subchannel): limit calling updateLLMConnection

16219de

chatgpt-codex-connector bot reviewed Apr 3, 2026

View reviewed changes

fix(subchannel): fire-and-forget admit token prefetch

105eb82

mickelr approved these changes Apr 3, 2026

View reviewed changes

Merge branch 'next' into feat/datachannel_in_lobby_register_and_disco…

9d20822

…nnect_reconnect_subscription

chatgpt-codex-connector bot reviewed Apr 3, 2026

View reviewed changes

JudyZhuHz approved these changes Apr 3, 2026

View reviewed changes

mickelr merged commit 36fd83f into webex:next Apr 3, 2026
21 of 22 checks passed

Conversation

Tianhui-Han commented Mar 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

COMPLETES #< INSERT LINK TO ISSUE >

This pull request addresses

by making the following changes

Change Type

The following scenarios were tested

The GAI Coding Policy And Copyright Annotation Best Practices

I certified that

Uh oh!

aws-amplify-us-east-2 bot commented Mar 30, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 31, 2026

Choose a reason for hiding this comment

Uh oh!

mickelr Apr 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

mickelr Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Apr 3, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot Apr 3, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Apr 3, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot commented Apr 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Tianhui-Han commented Mar 30, 2026 •

edited

Loading

mickelr Apr 2, 2026 •

edited

Loading