Data loss after docker restart/update

Questions
1

Describe the problem/error/question

Hey all,

I am sure some of you experienced this as well but I really want to check for the root cause of my data loss that meanwhile I found out happened multiple times already but I always that this was an error on my side. I really lost hours of work meanwhile due to this bug.

I am running n8n on a local server via docker run (portaner, not a stack) and map -v /opt/n8n:/home/node/.n8n

I can also see the sqlite database on my host machine and it seems like there are real “save points” from time to time which is then the time I get kicked back to.

With todays restart, I lost about 3 days of executions - but the last update was on Nov 5th, so for 3 days saving worked fine and then suddenly stopped which les tocms datacloss:

image
image778×178 11 KB

while this should be the latest stage:

image
image774×324 18.4 KB

I already accepted the circumstance that I have lost my data but I really want to prevent this happening with every reboot. Maybe someone has an idea

What is the error message (if any)?

I dont see any error

Please share your workflow

Not linked to workflow

Share the output returned by the last node

n/r

Information on your n8n setup

  • n8n version: 1.119.1
  • Database (default: SQLite): sqlite
  • n8n EXECUTIONS_PROCESS setting (default: own, main): n/r
  • Running n8n via (Docker, npm, n8n cloud, desktop app): Docker (run)/Portainer
  • Operating system: Ubuntu 24.04
2

To add on, I think the issue may be with a corrupted database

When I edited a Workflow I saw for a short time my tags I had assigned before the data loss but when assinging it, I got a foreign key error. Also in the tags management, I couldnt see the old tags and after a refresh, they were not selectable anymore - however it was a new tab so I would assume this couldn’t have been the cache

Really strange behavior :frowning:

3

hello @Wall-E

Why should the latest stage be earlier than the first screen?

And the difference in about 500 executions is quite strange. Did you miss any workflows, or only executions? As n8n cleans up old executions periodically.

4

I also lost lots of changes so workflows. Meanwhile I found out (after reading similar cases on reddit) that the issue is with commits to sqlite database and other users experienced exactly the same. I have changed to PostgreSQL meanwhile and never saw this happen again so far. So there must be issues with sqlite that users don’t see because (probably?) browser is caching these changes