Packet loss test

Find out whether repeated requests to a website drop or time out — with a plain-English verdict, not just numbers.

How this test works: a traditional packet loss test sends ICMP packets ("pings"). PingStuff instead sends 8 real web requests — the thing that actually matters for whether visitors can load your site — and reports how many complete.

Reading the results

A request counts as delivered when the server answers with any HTTP response within 8 seconds — even an error page proves the network carried the request and the reply. A request counts as dropped when the connection fails outright, and requests that take longer than 8 seconds to complete count as dropped too. The requests run one after another from a single PingStuff location, so this samples the path over time rather than hitting the site with a burst.

Want to know when requests start failing, not just whether they failed right now? Turn on monitoring and PingStuff checks your site on a schedule and alerts you when it stops answering.

Questions

How is this different from an ICMP packet loss test?
Classic packet loss tools send ICMP echo packets ("pings") and count how many come back. PingStuff doesn't send ICMP — and ICMP wouldn't tell you whether the website actually works, only whether the machine answers pings. Instead we send 8 real HTTP requests, the same kind a visitor's browser makes, and report how many complete. If these drop, your visitors feel it.
What causes dropped web requests?
An overloaded or crashing server, network congestion or a flaky route between networks, a struggling load balancer or reverse proxy, DNS trouble mid-request, or a firewall or rate limiter cutting connections. A one-off drop is often transient congestion; drops that repeat run after run usually point at the server or its network.
Is 1 of 8 requests failing bad?
One dropped request is 12.5% of this small sample — worth a second look, not an alarm. Run the test again: a single drop that never repeats was probably transient. A drop that shows up every run means roughly one in eight visitors is hitting a slow or failed page load, and that's worth fixing.
How do I monitor this over time?
An 8-request sample is a snapshot. PingStuff monitoring checks your site on a schedule around the clock, records failures and response times, and alerts you when requests start failing — so you catch trouble before your visitors report it.

Related tools

  • Latency Test Response time, phase by phase — multi-region rolling out.
  • Response Time Test How fast your site responds, broken down phase by phase.