fix(backend): Clean up orphaned schedules without schedule_id

[Bentlybro]

Changes 🏗️

Fixes AUTOGPT-SERVER-6W2 (~30K events) and AUTOGPT-SERVER-6W3 (~30K events) — combined 60K+ Sentry errors since Nov 17, 2025.

Problem

Old scheduled jobs created before schedule_id was added to GraphExecutionJobArgs have schedule_id=None. When these jobs fire and fail graph validation, _handle_graph_validation_error couldn't unschedule them because it only knew how to delete by schedule_id. Instead it logged:

Unable to unschedule graph: <id> as this is an old job with no associated schedule_id please remove manually

These jobs then kept firing on their cron schedule, failing validation, logging the error, and repeating — forever.

Why not use _cleanup_orphaned_schedules_for_graph?

The existing _cleanup_orphaned_schedules_for_graph() helper (used in _handle_graph_not_available) deletes all schedules for a given graph_id + user_id. That's fine when a graph is deleted/archived — all its schedules are invalid. But in the validation error case, only the old orphaned job failed. A user could have both an old legacy schedule (no schedule_id) and a newer valid schedule for the same graph. Using the broad cleanup would incorrectly nuke the valid one too.

Fix

New helper _cleanup_old_schedules_without_id() that:

1. Fetches all schedules for the graph_id + user_id
2. Skips any schedule where schedule_id is not None (newer, valid jobs)
3. Only deletes legacy jobs with schedule_id=None

This ensures old orphaned schedules get cleaned up automatically while preserving any valid newer schedules the user may have created.

1 file changed, 33 insertions, 2 deletions.

Impact

• Eliminates ~60K Sentry errors from 6W2/6W3
• Old orphaned schedules will self-clean on next validation failure
• No effect on new jobs (which always have schedule_id)
• Preserves valid newer schedules for the same graph (unlike the broader cleanup)

[qodo-code-review]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Cleanup semantics Confirm _cleanup_orphaned_schedules_for_graph(graph_id, user_id) is safe and correct in this validation-error context: it should be idempotent, tolerate “already deleted” schedules, and not remove unrelated schedules (e.g., multiple schedules for the same graph/user, if that’s possible). Also validate expected behavior if the cleanup call fails (retries vs. leaving the schedule to continue firing). logger.warning( f"Old scheduled job for graph {args.graph_id} has no schedule_id, " f"attempting cleanup by graph_id lookup" ) await _cleanup_orphaned_schedules_for_graph(args.graph_id, args.user_id) Logging quality The new warning message is an improvement, but consider including more context (e.g., job identifier / schedule name if available in args, and user_id) to make remediation and correlation easier while still avoiding noisy logs. logger.warning( f"Old scheduled job for graph {args.graph_id} has no schedule_id, " f"attempting cleanup by graph_id lookup" )

[coderabbitai]

Walkthrough

The scheduler's GraphValidationError handling was changed to call a new private helper, _cleanup_old_schedules_without_id, which queries and deletes schedules that lack a schedule_id and logs outcomes; when a schedule_id exists the existing delete path is retained and no public APIs were changed.

Changes

Cohort / File(s)	Summary
Scheduler - helper added & error handling updated autogpt_platform/backend/backend/executor/scheduler.py	Added private async helper _cleanup_old_schedules_without_id(graph_id: str, user_id: str) that finds and deletes schedules with schedule_id is None. Updated _handle_graph_validation_error to log a warning and invoke this helper for schedules missing schedule_id; existing deletion path remains when a schedule_id is present.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐰 I sniffed out schedules lost in the mist,
I nibbled the None-IDs off the list,
A gentle warning, then tidy and spry,
Old ghosts removed — hop, clean, and goodbye! 🥕✨

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 60.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title 'fix(backend): Clean up orphaned schedules without schedule_id' clearly and concisely describes the main change: fixing an issue by cleaning up orphaned schedules that lack a schedule_id.
Description check	✅ Passed	The PR description clearly relates to the changeset, detailing the problem with orphaned schedules and explaining the fix implemented in the scheduler code.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch fix/scheduler-orphaned-schedule-cleanup

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[ntindle]

Approved ✅

Disclosure: I'm an AI agent (Claude) acting on behalf of @ntindle.

Review Process

1. Human code review: @ntindle reviewed the code changes (per AutoGPT's humans-must-read-code policy)
2. Automated testing: I was asked to verify the changes and validate against Sentry data
3. Explicit approval direction: After reviewing my findings, @ntindle directed me to approve

Verification

Sentry issues confirmed:

• AUTOGPT-SERVER-6W2: 32,287 events — "Unable to unschedule graph... no associated schedule_id please remove manually"
• AUTOGPT-SERVER-6W3: 32,287 events — "Scheduled Graph... failed validation"
• Both still firing as of 5 minutes ago
• Combined: ~65K events since Nov 17, 2025

Code review:

• ✅ New _cleanup_old_schedules_without_id() correctly filters by schedule_id is None
• ✅ Preserves valid newer schedules (only removes legacy orphaned ones)
• ✅ Proper error handling with try/except and logging
• ✅ Uses correct ID for deletion (schedule.id, not schedule.schedule_id)
• ✅ Clear docstring explaining difference from existing cleanup function

Design rationale verified:

• Existing _cleanup_orphaned_schedules_for_graph() deletes ALL schedules (for deleted graphs)
• New function is surgical — only targets schedule_id=None jobs while preserving valid ones

Conclusion

This fix will eliminate ~65K recurring Sentry errors from orphaned schedules that have been firing since November 2025.

— Claude (AI agent, approved at @ntindle's direction)

[majdyz]

This is fine kind of, but what's blocking us on simply doing a one off cleanup on the database instead of doing this ?

[autogpt-reviewer]

PR #11927 — fix(backend): Clean up orphaned schedules without schedule_id

Author: Bentlybro | Files: scheduler.py (+33/−2) | CI: ✅ All green (tests 3.11/3.12/3.13, types, lint, CodeQL, security)

---

🎯 Verdict: APPROVE

---

What This PR Does

Adds a new helper _cleanup_old_schedules_without_id() that surgically removes legacy scheduled jobs (those with schedule_id=None) while preserving valid newer schedules. This fixes ~65K recurring Sentry errors (AUTOGPT-SERVER-6W2 + 6W3) from orphaned jobs created before schedule_id was added to GraphExecutionJobArgs. These jobs fire on cron, fail validation, log errors, and repeat — forever. The fix is self-healing: each orphan fires one more time, gets cleaned up, and never fires again.

---

Specialist Findings

🛡️ Security ✅ — No concerns. Authorization enforced at service layer (user_id check in delete_graph_execution_schedule). No cross-user deletion risk. Parameters come from internal job args, not external input. No info leakage (logs only).

🏗️ Architecture ✅ — New function is the right approach. Different invariants from existing _cleanup_orphaned_schedules_for_graph (which nukes ALL schedules for deleted graphs). Separation of concerns is clean. Minor duplication could be extracted to a shared helper, but tolerable at this scale.

⚡ Performance ✅ — Self-limiting cleanup path. Full jobstore scan is pre-existing (APScheduler API limitation, not introduced by this PR). Sequential deletes are acceptable given expected cardinality ≈1. Each orphan triggers cleanup at most once before being removed.

🧪 Testing ⚠️ — No tests included, but the entire scheduler module has zero test coverage (systemic issue). Function is simple enough for high confidence by inspection. Should add tests (mixed schedule filtering, deletion failure continuation) but not a merge blocker for this fix.

📖 Quality ✅ — Clean code. Good naming (matches _cleanup_* convention). Clear docstring explaining difference from sibling function. Correct log level change (error → warning). Error handling matches existing patterns (per-schedule try/except so one failure doesn't block others).

📦 Product ✅ — Safe for users. Deleted schedules were already permanently broken (fail validation on every fire, never execute successfully). Silent removal is appropriate — notifying users about forgotten legacy schedules would create more confusion than value. Self-healing approach is product-appropriate.

📬 Discussion ✅ — 2 approvals (ntindle, majdyz). majdyz's collateral deletion concern was correctly rebutted by Bentlybro (schedule.schedule_id vs schedule.id distinction). One-off DB cleanup question acknowledged but not blocking — majdyz approved despite it.

🔎 QA ✅ — Live testing passed. Frontend loads normally (landing, login, signup, dashboard, build page, marketplace). Backend healthy. Schedule Run button present and accessible. 7 screenshots captured. No regressions.

QA Screenshots:

• Landing
• Login
• Dashboard
• Tasks/Library
• Build Page
• Schedule Dialog
• Marketplace

---

Blockers

None.

Should Fix (Follow-up OK)

1. scheduler.py — Add unit tests for _cleanup_old_schedules_without_id: mixed schedule filtering, deletion failure continuation, empty schedule list. (Systemic gap — entire scheduler module is untested.)
2. scheduler.py:220-270 — Extract shared delete-iterate-log helper to reduce duplication between _cleanup_orphaned_schedules_for_graph and the new function.
3. Companion one-off DB cleanup — Consider a one-time DELETE against orphaned APScheduler jobs for immediate Sentry noise reduction. The runtime fix only cleans each orphan on its next cron fire, which could take hours/weeks depending on cron expressions.
4. scheduler.py:196 — Add user_id to warning message (partially done — already includes it, but qodo suggested more context like job identifier).

Risk Assessment

Merge risk: LOW | Rollback: EASY (revert removes cleanup; orphans resume failing harmlessly as before)

@ntindle Clean, well-scoped fix that eliminates ~65K Sentry errors. Self-healing approach is correct. No blockers — approve and merge.

[github-actions]

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

[CLAassistant]

All committers have signed the CLA.

[github-actions]

Conflicts have been resolved! 🎉 A maintainer will review the pull request shortly.

[github-actions]

🔍 PR Overlap Detection

This check compares your PR against all other open PRs targeting the same branch to detect potential merge conflicts early.

🟢 Low Risk — File Overlap Only

These PRs touch the same files but different sections (click to expand)

• fix(backend/copilot): dream runtime correctness — phase timeouts, ingestion drain, honest registration #13338 (ntindle · updated 34m ago)
  • Shared files: autogpt_platform/backend/backend/executor/scheduler.py

Summary: 0 conflict(s), 0 medium risk, 1 low risk (out of 1 PRs with file overlap)

---

Auto-generated on push. Ignores: openapi.json, lock files.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.38%. Comparing base (44a9d57) to head (12ae0be).
⚠️ Report is 4 commits behind head on dev.

Additional details and impacted files

@@            Coverage Diff             @@
##              dev   #11927      +/-   ##
==========================================
+ Coverage   76.33%   76.38%   +0.04%     
==========================================
  Files        2701     2703       +2     
  Lines      206224   206911     +687     
  Branches    19777    19818      +41     
==========================================
+ Hits       157417   158039     +622     
- Misses      44444    44508      +64     
- Partials     4363     4364       +1     

Flag	Coverage Δ
platform-backend	83.15% <100.00%> (+0.04%)	⬆️
platform-frontend-e2e	30.93% <ø> (-0.22%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
Platform Backend	83.15% <100.00%> (+0.04%)	⬆️
Platform Frontend	50.21% <ø> (-0.08%)	⬇️
AutoGPT Libs	∅ <ø> (∅)
Classic AutoGPT	28.43% <ø> (ø)

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[sentry]

Issues attributed to commits in this pull request

This pull request was merged and Sentry observed the following issues:

• ‼️ RuntimeError: boom in app:local-behave:local