Hi,

I got the Active CPU warning email and wanted to explain what happened before it becomes a problem, since the cause was me.

Between 15 and 18 August I ran load tests against my own production domain, which in hindsight was a bad idea. I exported my logs afterwards and it's pretty clear — in one seven minute window on the 15th, 49,118 out of 49,342 requests were from "Grafana k6/2.2.0". That run alone caused 11,400 gateway timeouts and one of my API routes was averaging 92 seconds a call because the test had flattened my database.

While digging into it I also found a real bug of mine. Next.js prefetches Link by default and I hadn't turned it off on my product cards, so just scrolling the catalog was server rendering every product page in view. Your per-route usage page made it obvious: 4,200 invocations of /[locale]/catalog/[id] versus 60 of /api/products/[id]/view, and that second one only fires when someone actually opens a product. So around 4,140 page renders for nobody.

Both are fixed and deployed. I deleted the whole k6 suite from the repo and the test accounts with it, so I can't accidentally point a load test at production again, and prefetching is off on the product cards and the sidebar.

What I'm hoping to find out is whether I'm about to get paused, and whether anything can be done given the spike is historical and the cause is gone. The project is still in testing and the usage from here should be a fraction of what those days show. I can share the log export or the commits if that helps.

Team: assayo95s-projects Project: sourcely-dz (www.sourcelydz.com)

Thanks