Workflows, credentials and execution data lost

[David_A_T.C]

I’m experiencing a critical data loss issue with my n8n installation where workflows and their execution history aswell as credentials have mysteriously disappeared from the database over a specific time period.

Environment

• n8n version: 1.120.3 (happened on older version)

• Installation type: Docker

• Database: SQLite

• Database location: /var/lib/docker/volumes/n8n_data/_data/database.sqlite

• OS: Ubuntu

• VPS Provider: Hostinger

What Happened

1. Missing Data Period: Workflows and credentials created during a specific period of several weeks have completely disappeared

2. Older Workflows Intact: Workflows created before this period are still present and working fine

3. Execution History Lost: The execution history for the missing workflows is also gone

4. Not Related to Update: Initially suspected an n8n update, but after restoring a backup from before the update, the workflows were already missing in that backup too

Database Investigation

Using DBeaver to connect to the SQLite database, I confirmed:

• The workflow_entity table shows a clear gap in the createdAt dates

• The execution_entity table is missing corresponding execution records

• No soft-delete flags or similar indicators that would suggest recoverable data

• The database file itself shows no obvious corruption

What I’ve Tried

1. Checked multiple backup points - workflows already missing in all of them

2. Verified database integrity - no corruption detected

3. Confirmed no accidental manual deletion (no one else has access)

Questions

1. Has anyone experienced similar data loss with SQLite-based n8n installations?

2. Are there known issues with SQLite data integrity under specific conditions?

3. Could this be related to:

   • Docker volume mounting issues?

   • SQLite WAL (Write-Ahead Logging) mode problems?

   • Disk space issues causing incomplete writes?

   • File system issues on the VPS?

4. Is there any way to recover deleted rows from SQLite? (I still have access to the database file)

Additional Context

• The missing workflows were under test and inactive, executing successfully before they disappeared

• No disk space issues were present during the affected period

• Server uptime was normal with no unexpected restarts

• No backup system was configured at the time (now implementing automated backups)Request for Help

  Has anyone experienced something similar? Any suggestions on:

  • Potential root causes to investigate?

  • Recovery methods for SQLite databases?

  • Best practices to prevent this from happening again?

  • Should I migrate from SQLite to PostgreSQL for better reliability?

  Any insights would be greatly appreciated. This is a production environment and losing workflows is critical for our operations. Thanks for your support

[comfdy]

following this up cos i experienced similar data loss with SQLite-based n8n installation

[Parintele_Damaskin]

@David_A_T.C

• Disk space issues causing incomplete writes?
• File system issues on the VPS?

Same I thought when I was arguing with h.s.tngeer support, saying that is everything OK, but my corrupted db wasn’t!

Lost 2 days of intensive work…

They suggested as well to use FQLite (SQLite Forensic Toolkit) or the SQLite CLI to check corrupt files.

@David_A_T.C

Should I migrate from SQLite to PostgreSQL for better reliability?

Highly RECOMMENDED!

@David_A_T.C

Any insights would be greatly appreciated.

Yea… just backup the .n8n with a bash script(sometimes I do it hourly .