Warps samples latency, jitter and packet loss hop by hop, then keeps a rolling history so you can tell a bad minute from a bad route.
Get startedRead the docsp50/p95/p99 over a sliding window instead of a single ping. Slow tails show up before users complain.
Correlates loss with the hop that introduced it, so transit problems are not blamed on the destination.
A single static binary, roughly 9 MB, no database required for the default collector.
Point it at a host and let it run. Output is line-delimited JSON by default.
$ warps probe --target example.net --interval 5s --window 15m
{"t":"2026-09-06T18:20:05Z","target":"example.net","rtt_p50":11.4,"rtt_p95":18.9,"loss":0.00}
{"t":"2026-09-06T18:20:10Z","target":"example.net","rtt_p50":11.6,"rtt_p95":19.4,"loss":0.00}
| Region | p50 | p95 | Loss | State |
|---|---|---|---|---|
| eu-central | 11.4 ms | 18.9 ms | 0.00 % | healthy |
| eu-west | 19.2 ms | 27.7 ms | 0.00 % | healthy |
| us-east | 88.6 ms | 104.1 ms | 0.02 % | healthy |
| ap-southeast | 163.0 ms | 191.5 ms | 0.05 % | healthy |