I’m running the Cloud version 2.17.6 and keep getting these “connection lost” errors. The status in the upper right hand of the screen keeps changing from ready to “connection lost” and even when it’s functional I get “connection lost” mid-execution. My internet connection seems perfectly stable for every other use case
Attaching a screenshot of the workflow as well as the overall status errors.
Hi @terraxus have you restarted your n8n instance from the cloud admin dashboard?
Hi @terraxus, that “Lost connection to the server” mid-execution usually means your cloud instance is running out of memory and crashing — check your execution list for runs that just vanish with no output, that confirms it. if your workflow processes large payloads, add a Split In Batches node before the heavy part to keep memory usage down. what kind of data volume are you pushing through per execution?
@Anshul_Namdev thanks for the suggestion, tried it but it didn’t help
@achamm there were no concurrent executions running at that time nor do we deal with large payloads unfortunately
@terraxus tried downgrading to a version if available? As if you have tried that then there are almost no option available on the client side to tweak. And if the problem persists for 3-4 days then you can reach out to the support.
It seems to be a global issue, this is what I got back from support
Thank you for the detailed report. We have identified the root cause of the connection drops you are experiencing.
There are two scenarios that trigger the issue:
- A user opens n8n in multiple tabs within the same browser. Both tabs end up sharing the same internal connection identifier, so each tab’s connection immediately terminates the other’s.
- Multiple users working on the same workflow at the same time. The real-time collaboration handling introduced in 2.16.0 causes the same conflict, resulting in the same reconnect loop.
In the meantime, here is the workaround:
- Keep only one n8n tab open per browser per user.
- Avoid having multiple users editing the same workflow simultaneously until the fix is out. Users can view workflows without issue, but concurrent editing will trigger the disconnects.
- If a user needs a second session, opening n8n in a separate browser or an incognito window works without issue.
We have prioritized a fix internally and expect it to ship within the next 4 days. Once it is out, there will be no action required on your end.
I’m experiencing the exact same issue. Ever since I updated to version 2.17.5, this problem has been persistent. I’ve found very little information online, and the few potential fixes I did come across don’t seem to work.
Have you tried the “one tab at a time” solution? It worked for me. I still have to update my instance to figure out if the newer version helped
