Usage: include reset and deleted session archives by rcrick · Pull Request #43215 · openclaw/openclaw

rcrick · 2026-03-11T13:34:32Z

Summary

Problem: Usage aggregation only counted active *.jsonl session transcripts and skipped archived *.jsonl.reset.* / *.jsonl.deleted.* transcripts.
Why it matters: Usage page totals become much lower than actual spend after /new, session reset, or deleting old sessions.
What changed: usage scanning and session discovery now include reset and deleted transcript archives, and per-session usage detail falls back to archived transcripts when the active file no longer exists.
What did NOT change (scope boundary): *.jsonl.bak.* archives are still excluded to avoid double-counting backup copies.

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

Closes Bug: Usage page doesn't count tokens from reset and delete sessions #42032
Related #

User-visible / Behavior Changes

Usage totals now include spend from archived reset and deleted session transcripts.
Session usage listing can discover archived reset/deleted sessions.
Per-session usage detail can still load logs/timeseries/summary after the active transcript has been rotated away.

Security Impact (required)

New permissions/capabilities? (No)
Secrets/tokens handling changed? (No)
New/changed network calls? (No)
Command/tool execution surface changed? (No)
Data access scope changed? (No)
If any Yes, explain risk + mitigation:

Repro + Verification

Environment

OS: macOS
Runtime/container: Node 24 / pnpm
Model/provider: N/A
Integration/channel (if any): Control UI usage aggregation
Relevant config (redacted): OPENCLAW_STATE_DIR pointed at temp test state dirs

Steps

Create session transcripts for active *.jsonl, archived *.jsonl.reset.*, archived *.jsonl.deleted.*, and backup *.jsonl.bak.*.
Run usage aggregation and session discovery against that state directory.
Query per-session usage detail for a session that only exists as an archived reset transcript.

Expected

Active, reset, and deleted transcripts are counted in usage totals.
Backup bak archives are not counted.
Per-session usage detail still works when only the archived transcript remains.

Actual

Before this change, only active *.jsonl transcripts were counted, so reset/deleted usage was omitted.

Evidence

Failing test/log before + passing after
Trace/log snippets
Screenshot/recording
Perf numbers (if relevant)

Targeted tests run:

corepack pnpm vitest run src/config/sessions/artifacts.test.ts src/infra/session-cost-usage.test.ts src/gateway/server-methods/usage.sessions-usage.test.ts

Result: 23 tests passed.

Human Verification (required)

What you personally verified (not just CI), and how:

Verified scenarios: counted active/reset/deleted transcripts in aggregate usage; excluded bak; loaded summary/timeseries/logs from archived reset transcript fallback.
Edge cases checked: archived session discovery still derives the original session id; specific session usage lookup still resolves when the active file is gone.
What you did not verify: full manual Control UI click-through against a live local gateway.

Review Conversations

I replied to or resolved every bot review conversation I addressed in this PR.
I left unresolved only the conversations that still need reviewer or maintainer judgment.

If a bot review conversation is addressed by this PR, resolve that conversation yourself. Do not leave bot review conversation cleanup for maintainers.

Compatibility / Migration

Backward compatible? (Yes)
Config/env changes? (No)
Migration needed? (No)
If yes, exact upgrade steps:

Failure Recovery (if this breaks)

How to disable/revert this change quickly: revert this PR/commit to restore the previous active-transcript-only behavior.
Files/config to restore: src/config/sessions/artifacts.ts, src/infra/session-cost-usage.ts, src/gateway/server-methods/usage.ts
Known bad symptoms reviewers should watch for: usage totals unexpectedly jumping from counting backup archives, or archived session detail lookups returning duplicate/incorrect sessions.

Risks and Mitigations

Risk: counting archived files could accidentally include backup copies and double-count usage.
- Mitigation: only reset and deleted archives are included; bak remains excluded, with test coverage for that boundary.

greptile-apps · 2026-03-11T13:37:28Z

Greptile Summary

This PR fixes a real under-counting bug: usage aggregation and session discovery previously ignored *.jsonl.reset.* and *.jsonl.deleted.* archive transcripts, causing the Usage page to show totals much lower than actual spend after a session reset or deletion. The fix is well-scoped — *.jsonl.bak.* files remain excluded to prevent double-counting — and the new helper functions (isUsageCountedSessionTranscriptFileName, parseUsageCountedSessionIdFromFileName) are clean and well-tested.

Two logic issues need attention before merging:

Duplicate sessions in list view after a reset (src/infra/session-cost-usage.ts, discoverAllSessions): After /reset, both sess.jsonl (new active file) and sess.jsonl.reset.<TIMESTAMP> (archived pre-reset file) exist on disk simultaneously. discoverAllSessions now emits both as separate DiscoveredSession entries with the same sessionId, and no downstream deduplication occurs, so the same session appears twice in the UI session list.
Wrong archive selected when both reset and deleted archives exist (resolveExistingUsageSessionFile, line 335–338): The "latest archive" is chosen by a full-filename lexicographic sort. Because 'r' > 'd' in ASCII, a .reset. archive always sorts above a .deleted. archive, even when the deleted archive has a later timestamp (e.g., reset at T11:00, deleted at T12:00). The comparator should sort only on the timestamp suffix.

Confidence Score: 3/5

PR introduces two logic bugs — duplicate session entries in list view after reset, and incorrect archive selection when reset+deleted archives coexist — that should be addressed before merging.
The core concept and implementation are sound, and the new artifact helpers are correct. However, two logic bugs were found: (1) discoverAllSessions emits duplicate sessionId entries when an active file and a reset archive coexist, causing UI duplicates; (2) resolveExistingUsageSessionFile's lexicographic sort picks a reset archive over a newer deleted archive. Neither bug was caught by the new tests, which don't exercise the coexistence scenario. Score of 3 reflects meaningful but fixable issues that affect correctness in real post-reset workflows.
src/infra/session-cost-usage.ts — specifically discoverAllSessions (deduplication by sessionId) and resolveExistingUsageSessionFile (archive sort comparator).

Comments Outside Diff (1)

src/infra/session-cost-usage.ts, line 449-517 (link)

Duplicate session IDs when active transcript and reset archive coexist

After a /reset, the filesystem holds both sess.jsonl (the new, post-reset active file) and sess.jsonl.reset.<TIMESTAMP> (the archived pre-reset transcript). discoverAllSessions now includes both files, and parseUsageCountedSessionIdFromFileName maps them to the same sessionId ("sess"). Because there is no deduplication by sessionId inside this loop, the result contains two DiscoveredSession entries with identical sessionId values.

In discoverAllSessionsForUsage (also not deduplicated) and then in the usage.ts list-view merge loop, both entries survive and are emitted to mergedEntries, causing the same session to appear twice in the UI session list.

The test added in this PR only covers the case where only the archived transcript exists (no active file), so it doesn't catch this scenario.

A fix would be to deduplicate by sessionId in discoverAllSessions, keeping the entry whose sessionFile is the most recently modified:

// After building `discovered`, deduplicate by sessionId keeping newest mtime
const bySessionId = new Map<string, DiscoveredSession>();
for (const entry of discovered) {
  const existing = bySessionId.get(entry.sessionId);
  if (!existing || entry.mtime > existing.mtime) {
    bySessionId.set(entry.sessionId, entry);
  }
}
return [...bySessionId.values()].toSorted((a, b) => b.mtime - a.mtime);

Prompt To Fix With AI

This is a comment left during a code review.
Path: src/infra/session-cost-usage.ts
Line: 449-517

Comment:
**Duplicate session IDs when active transcript and reset archive coexist**

After a `/reset`, the filesystem holds *both* `sess.jsonl` (the new, post-reset active file) **and** `sess.jsonl.reset.<TIMESTAMP>` (the archived pre-reset transcript). `discoverAllSessions` now includes both files, and `parseUsageCountedSessionIdFromFileName` maps them to the same `sessionId` ("sess"). Because there is no deduplication by `sessionId` inside this loop, the result contains two `DiscoveredSession` entries with identical `sessionId` values.

In `discoverAllSessionsForUsage` (also not deduplicated) and then in the `usage.ts` list-view merge loop, both entries survive and are emitted to `mergedEntries`, causing the same session to appear twice in the UI session list.

The test added in this PR only covers the case where *only* the archived transcript exists (no active file), so it doesn't catch this scenario.

A fix would be to deduplicate by `sessionId` in `discoverAllSessions`, keeping the entry whose `sessionFile` is the most recently modified:

```ts
// After building `discovered`, deduplicate by sessionId keeping newest mtime
const bySessionId = new Map<string, DiscoveredSession>();
for (const entry of discovered) {
  const existing = bySessionId.get(entry.sessionId);
  if (!existing || entry.mtime > existing.mtime) {
    bySessionId.set(entry.sessionId, entry);
  }
}
return [...bySessionId.values()].toSorted((a, b) => b.mtime - a.mtime);
```

How can I resolve this? If you propose a fix, please make it concise.

_{Last reviewed commit: 007b637}

src/infra/session-cost-usage.ts

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 007b6378db

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".

src/infra/session-cost-usage.ts

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 79a501581f

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".

src/infra/session-cost-usage.ts

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: afa5239c1d

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".

src/infra/session-cost-usage.ts

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b167ed4975

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 · 2026-03-11T15:49:50Z

src/infra/session-cost-usage.ts


  for (const entry of entries) {
-    if (!entry.isFile() || !entry.name.endsWith(".jsonl")) {
+    if (!entry.isFile() || !isUsageCountedSessionTranscriptFileName(entry.name)) {


Filter archived sessions by transcript activity window

Including *.jsonl.reset.*/*.jsonl.deleted.* in discovery here makes the existing mtime-based date filter inaccurate for archived transcripts, because archive rename time is often much newer than the messages inside the file. If a user deletes an old session today and then requests a recent range (for example, last 7 days), that archive is discovered even when it has no in-range activity; sessions.usage applies limit before loading per-session usage, so these zero-usage rows can displace real in-range sessions. This path should gate archive inclusion on transcript timestamps (or drop sessions with no in-range usage before limiting).

Useful? React with 👍 / 👎.

frankekn · 2026-03-23T04:46:55Z

@codex review

chatgpt-codex-connector · 2026-03-23T04:54:43Z

Codex Review: Didn't find any major issues. Swish!

ℹ️ About Codex in GitHub

Your team has set up Codex to 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 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

frankekn · 2026-03-23T05:10:29Z

Merged via squash.

Prepared head SHA: 49ed6c2fa30ff512b55f680a049a9a316175f373
Merge commit: 2fe1ff8ea8ab86473d9c69784d2f1c80939e3710

Thanks @rcrick!

@frankekn

Merged via squash. Prepared head SHA: 49ed6c2 Co-authored-by: rcrick <23069968+rcrick@users.noreply.github.com> Co-authored-by: frankekn <4488090+frankekn@users.noreply.github.com> Reviewed-by: @frankekn

openclaw-barnacle bot added gateway Gateway runtime size: M labels Mar 11, 2026

greptile-apps bot reviewed Mar 11, 2026

View reviewed changes

src/infra/session-cost-usage.ts Show resolved Hide resolved

chatgpt-codex-connector bot reviewed Mar 11, 2026

View reviewed changes

src/infra/session-cost-usage.ts Outdated Show resolved Hide resolved

src/infra/session-cost-usage.ts Show resolved Hide resolved

chatgpt-codex-connector bot reviewed Mar 11, 2026

View reviewed changes

src/infra/session-cost-usage.ts Outdated Show resolved Hide resolved

rcrick force-pushed the fix/42032-usage-include-archived-sessions branch 2 times, most recently from 943fe08 to afa5239 Compare March 11, 2026 15:01

chatgpt-codex-connector bot reviewed Mar 11, 2026

View reviewed changes

src/infra/session-cost-usage.ts Outdated Show resolved Hide resolved

chatgpt-codex-connector bot reviewed Mar 11, 2026

View reviewed changes

openclaw-barnacle bot added size: L and removed size: M labels Mar 11, 2026

frankekn force-pushed the fix/42032-usage-include-archived-sessions branch from b167ed4 to 5cf4c67 Compare March 23, 2026 04:42

rcrick and others added 5 commits March 23, 2026 12:42

Usage: include reset and deleted session archives

8fd9f81

Usage: fix archived session fallback ordering

1ec27e4

Usage: prefer active sessions during archive dedupe

b4d1944

Usage: honor custom archive directories in fallback

3caa9b4

fix(usage): count archived reset and deleted transcripts

49ed6c2

frankekn force-pushed the fix/42032-usage-include-archived-sessions branch from 5cf4c67 to 49ed6c2 Compare March 23, 2026 04:42

frankekn self-assigned this Mar 23, 2026

frankekn merged commit 2fe1ff8 into openclaw:main Mar 23, 2026
37 of 38 checks passed

Uh oh!

Conversation

rcrick commented Mar 11, 2026

Summary

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

User-visible / Behavior Changes

Security Impact (required)

Repro + Verification

Environment

Steps

Expected

Actual

Evidence

Human Verification (required)

Review Conversations

Compatibility / Migration

Failure Recovery (if this breaks)

Risks and Mitigations

Uh oh!

greptile-apps bot commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 3/5

Comments Outside Diff (1)

Uh oh!

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

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 11, 2026

Choose a reason for hiding this comment

Uh oh!

frankekn commented Mar 23, 2026

Uh oh!

chatgpt-codex-connector bot commented Mar 23, 2026

Uh oh!

Uh oh!

frankekn commented Mar 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

greptile-apps bot commented Mar 11, 2026 •

edited

Loading