source-tiktok-marketing: TikTok API returning code 50000 "System error." on /ad/get/ (2026-08-11)
- Dominant language
- Python
- Stars
- 22.1k
- Forks
- 5.3k
- PR merge metrics
- PR metrics pending
Description
## Summary
Since 2026-08-11 ~08:32 UTC, `source-tiktok-marketing` syncs have been failing with a TikTok-side generic internal error on `/open_api/v1.3/ad/get/`:
```
HTTP 200 {'code': 50000, 'message': 'System error.', 'request_id': '...', 'data': {}}
```
Sentry group: https://sentry.io/organizations/airbytehq/issues/7379973626/
## Impact
16 Cloud connections across 15 workspaces. Failures hit both unpinned actors (Cloud GA 5.1.5) and rollout-pinned actors (5.1.8/5.1.9), so this is a TikTok-side issue, not a connector regression. Reproduction with Airbyte integration-test credentials returns `code: 0 / OK`, so the endpoint/param shape itself is fine.
## Side effect on progressive rollouts
These TikTok failures tripped the AutoPilot failure threshold and auto-cancelled the 5.1.8 and 5.1.9 rollouts. Four affected actors remain pinned to 5.1.9 by cancelled rollout `faf00bdb-3e5f-4219-9fce-75e716264681`:
- `06b6f216-35fb-4e7c-9745-c65b4c72f62f` (apex-dev)
- `6d172e68-1f67-4516-9c1d-331bde692559`
- `da077cb0-b13b-4cfc-96ce-c68ebb679583`
- `a993d5a6-8e87-4bd7-ad66-fea54a1034b9`
## Action
Unpin those four actors so they return to Cloud GA 5.1.5, and monitor whether the TikTok 50000 burst subsides on its own.
---
**Internal Tracking:** https://github.com/airbytehq/oncall/issues/13272
Contributor guide
Assessment
This issue has not been assessed yet.