Workflow not found

[A_Z]

Workflows periodically disappear from n8n UI showing “Workflow not found” error, even though they exist in the PostgreSQL database. After restarting the n8n container, workflows become accessible again temporarily, but the issue recurs.

Timeline:

• Workflows created and work normally

• After some time (several hours), accessing workflow by URL shows “Workflow not found”

• Workflows are present in workflow_entity table in database

• Restarting n8n container temporarily fixes the issue

• Problem repeats periodically

Database verification:

SELECT id, name, active FROM workflow_entity WHERE name = ‘Публикация в Дзен’;
– Returns: id=‘itoFOJCyDizr4iah’, name=‘Публикация в Дзен’, active=false

What is the error message (if any)?

Workflow not found
We couldn’t find the workflow you were looking for. Make sure you have the correct URL.

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

Share the output returned by the last node

information on your n8n setup

• n8n version: 1.120.4

• Database: PostgreSQL 16.11

• n8n EXECUTIONS_PROCESS setting: own, main (default)

• Running n8n via: Docker

• Operating system: Ubuntu 24.04.2 LTS (Linux)

Additional context:

• n8n runs in Docker Compose setup

• PostgreSQL runs in separate container (self-hosted-n8n-postgres-1)

• Multiple workflows affected

[A_Z]

SOLUTION: n8n “Workflow not found” Error - Corrupted PostgreSQL Index

========================================

PROBLEM

Workflows periodically disappeared from n8n UI showing “Workflow not found” error, even though they existed in PostgreSQL database. Restarting n8n temporarily fixed it, but problem returned daily.

========================================

ROOT CAUSE

CORRUPTED B-TREE INDEX on workflow_entity.id column. The index returned 0 rows while actual data existed in the table.

========================================

DIAGNOSIS

Key test that revealed the issue:

SQL with index (BROKEN - returned 0 rows):
SELECT id, name FROM workflow_entity WHERE id = ‘your-workflow-id’;

SQL without index (WORKED - returned 1 row):
SET enable_indexscan = OFF;
SET enable_bitmapscan = OFF;
SELECT id, name FROM workflow_entity WHERE id = ‘your-workflow-id’;
RESET enable_indexscan;
RESET enable_bitmapscan;

Result: Sequential scan found data, but index scan didn’t = CORRUPTED INDEX

Additional findings:

• Two duplicate indexes on same id column (workflow_entity_pkey and pk_workflow_entity_id)
• High update frequency: 3,360 updates vs 88 inserts (38:1 ratio)
• Index actively used: 422,716 scans on corrupted index

========================================

IMMEDIATE FIX

Step 1: Rebuild indexes

REINDEX TABLE workflow_entity;
REINDEX TABLE shared_workflow;

Step 2: Restart n8n

docker restart n8n-container-name

Step 3: Verify fix

SELECT id, name FROM workflow_entity WHERE id = 'your-workflow-id';
(should now return 1 row)

========================================

LONG-TERM PREVENTION

Create automated nightly REINDEX via cron:

1. Create script /root/n8n-reindex.sh:

   #!/bin/bash
   LOG_FILE=“/var/log/n8n-reindex.log”
   echo “[$(date ‘+%Y-%m-%d %H:%M:%S’)] Starting REINDEX…” >> $LOG_FILE

   docker exec -i postgres-container-name psql -U postgres-user -d n8n -c “REINDEX TABLE workflow_entity;” >> $LOG_FILE 2>&1
   docker exec -i postgres-container-name psql -U postgres-user -d n8n -c “REINDEX TABLE shared_workflow;” >> $LOG_FILE 2>&1

   echo “[$(date ‘+%Y-%m-%d %H:%M:%S’)] REINDEX completed” >> $LOG_FILE

2. Make executable:

   chmod +x /root/n8n-reindex.sh

3. Add to crontab (runs daily at 3:00 AM UTC):

   crontab -e
   0 3 * * * /root/n8n-reindex.sh

========================================

WHY IT HAPPENED

Corruption likely caused by:

• High update frequency (38 updates per insert)
• Duplicate indexes competing for same data
• Possible improper PostgreSQL shutdowns (Docker restarts)
• Disk I/O issues

========================================

HOW TO TEST IF YOU HAVE THIS ISSUE

Run this SQL to test for index corruption:

SET enable_indexscan = OFF;
SET enable_bitmapscan = OFF;
SELECT id, name FROM workflow_entity WHERE id = 'your-missing-workflow-id';
RESET enable_indexscan;
RESET enable_bitmapscan;

If this returns a row but normal SELECT doesn’t = YOU HAVE INDEX CORRUPTION

========================================

SYSTEM INFO

• n8n version: 1.120.4
• Database: PostgreSQL 16.11
• Deployment: Docker Compose
• OS: Ubuntu 24.04.2 LTS

========================================

KEY TAKEAWAYS

1. Index corruption can be SILENT - data exists but queries return empty
2. Sequential scans bypass corrupted indexes (useful for diagnosis)
3. Regular REINDEX prevents corruption accumulation
4. High update frequency stresses indexes more than inserts
5. Duplicate indexes can cause conflicts under heavy load

Hope this helps others with similar issues!