Any ideas where I should look? I am running a self-hosted instance with 1 main and 5 workers. When I run a simple AI Agent workflow with local Redis memory and a few tools, everything works great. When I run the chat using the public URL, the agent works excellent but suddenly, after a few back and forth messages, starts over as if the conversation never happened. I have the Redis memory set to 0 for TTL and 20 for the context window. Only about 10 back and forth messages have been done. I checked the execution logs and the Redis key did not change at any time, but the conversation history returned to the AI Assistant at the start of a new message (after a few successful messages) suddenly becomes blank (all history gone). Any ideas what could be going on?
What is the error message (if any)?
No error
Please share your workflow
Share the output returned by the last node
Difficult to show. Output works but then just forgets previous context.
Information on your n8n setup
- n8n version: 1.95.2
- Database (default: SQLite): Postgres
- n8n EXECUTIONS_PROCESS setting (default: own, main): main
- Running n8n via (Docker, npm, n8n cloud, desktop app): docker
- Operating system: Debian 12