Urgent: Zeabur AI Agent auto-repair caused PostgreSQL data loss, requesting disk volume snapshot recovery
Account Email: [email protected] Account Phone: +86 18682458296
Project ID: project-697899a81eedf756b36ee5a5
PostgreSQL Service ID: 697899a91eedf756b36ee5a8 Region: Jakarta, Indonesia
Problem Overview:
Between 03:09-03:18 UTC on 2026-03-28, the Zeabur AI Agent deleted critical data files in the data volume while automatically repairing PostgreSQL, resulting in the loss of all N8N workflows, credentials, and other data.
Detailed Timeline (from runtime logs):
- 03:08 UTC — PostgreSQL running normally, last successful checkpoint.
- 03:09-03:16 UTC — Started seeing "Role 'postgres' does not exist" errors (suspected Agent intervention).
- 03:17:54 UTC — FATAL: could not open file "global/pg_filenode.map": No such file or directory (system files deleted).
- 03:18:16 UTC — Agent pulled new postgres:17 image, initdb failed due to non-empty directory, entered crash loop.
- 03:23:58 UTC — Agent created pgdata_new directory and initialized an empty database.
- Continued crashing and restarting since then.
Current Data Volume Status (/var/lib/postgresql/data/pgdata/):
- base/ — Does not exist (this directory contains all table data).
- global/ — Only a few files remain, pg_control is lost.
- pg_wal/ — Exists.
- PG_VERSION — Exists (manually restored).
- Other directories (pg_xact/, pg_multixact/, etc.) — All lost.
Request:
Please check if there is a disk volume snapshot for this PostgreSQL service available before 2026-03-28 03:08 UTC that can be restored. The base/ directory stores all N8N workflows and encrypted credentials, which are core business data.
Note: I have upgraded to a paid membership (upgraded on 2026-03-27). Please let me know if further log files are needed.