Service fails to start — node egress network i/o timeout (cannot pull image / reach any external service)

Hi Zeabur team,

My n8n service stopped starting after a restart. It keeps failing to pull the Docker image with
a dial tcp ... i/o timeout error, and it now cannot come back online.

Service info:

Project ID: 68ae5e9966a2f3f41ecc50ab
Service ID: 68ae5f4398a30c735c661ece
Environment ID: 68ae5e99f0ea38792a8660de
Domain: voccomputer.zeabur.app
Image: n8nio/n8n:2.6.3

What I've already ruled out:

Image is valid — n8nio/n8n:2.6.3 exists on Docker Hub and was successfully pulled earlier
today (last pull ~01:25 UTC). So this is not a missing/corrupted image.
Not a Docker Hub–specific issue — I changed the image registry to ghcr.io/n8n-io/n8n:2.6.3,
and it fails with the same timeout.
Not a runtime/credential issue — before the restart, the running container was also throwing
ETIMEDOUT when connecting to the Google Sheets API (216.239.34.223:443).

Conclusion:
Three different external targets over three different network paths (Docker Hub, GHCR, Google
API) all time out. This strongly indicates the node's outbound (egress) network is broken, not
an image or configuration problem on my side.

Redeploying does not help.

Request:
Could you please reschedule my service onto a healthy node, or investigate the egress/network
issue on the node currently hosting it? This is a production service and is currently down.

Thanks!

Hi, thanks for the detailed troubleshooting — your diagnosis was spot on.

We confirmed the node your service was scheduled on had a broken outbound network, which caused the image pull timeouts (and the earlier ETIMEDOUT errors to the Google Sheets API). This was a node-level issue on our side, not anything wrong with your configuration.

Your service has since been rescheduled onto a healthy node: the image pulled successfully, the container is running, and https://voccomputer.zeabur.app/healthz is returning 200. We're also taking the faulty node out of rotation to prevent recurrence.

Could you check on your side that everything (including your Google Sheets workflows) is working as expected?

Forum

1 Reply