Tracking only works while the system behind it is running. We host it on a secure server that repairs itself when something goes wrong, so your data keeps flowing even when nobody is watching.
If tracking breaks, your website still loads and your ads still run, so nothing looks wrong. We learned this first-hand in our lab: one bad request knocked our tracking server offline for four weeks without a single warning. That is why every system we hand over can recover by itself, and why we test that it does.
What it costs you: weeks of ad spend with no data to show for it.
Live walkthrough from the Limba Labs Engineering Lab · Open on YouTube ↗
The fix for the original crash has two parts: a restart policy so the container comes back, and an Nginx rule that turns the bad request away before it reaches the tracking server.
# Nginx: reject the malformed request before it reaches the container if ($request_uri ~ "^//") { return 400; } $ curl -si https://<tracking-domain>// HTTP/1.1 400 Bad Request Server: nginx/1.28.3 $ docker inspect -f 'Restarts: {{.RestartCount}}' sst-server Restarts: 1
Excerpt from the recorded terminal session. The domain is hidden.
We can check your current setup and tell you what would happen if it went down today.