Traceroute and MTR show the route from a check server to a host: every router (hop) your packets pass through. They help to find where a connection slows down or breaks. Both are available on check-host.org from 25 locations: traceroute and MTR.
Traceroute: the route
Each line is one hop. You see its number, the address of the router and the response time for each probe.
- Response time grows along the route — this is normal: the further the router, the longer the way.
- A sharp jump in time between two hops — usually a long physical distance, for example a link across an ocean.
*instead of a time — the router did not answer this probe. Many routers do not reply to traceroute on purpose; if the next hops answer, everything is fine.- Not reached — the last hops do not answer. The host may block probes, or the route really breaks there.
MTR: traceroute with statistics
MTR sends several probes to every hop and shows statistics:
- Loss — the share of probes without an answer;
- Sent — how many probes were sent;
- Last, Avg, Best, Worst — the last, average, best and worst response time;
- StDev — how much the time jumps; a high value means an unstable link.
Loss on a middle hop is often not a problem
Routers handle traffic passing through them faster than packets addressed to themselves. That is why a router can show 30–50% loss while all following hops show 0%. Such loss is not real: the router simply answers probes with a low priority.
Real loss looks different: it starts on some hop and continues on every hop after it, up to the destination. The network where it starts is the place to look for the problem.
What to do with the result
- Loss and growing times only on the last hop — the problem is on the destination server or its network.
- Loss starting in the middle and continuing to the end — a problem at the provider of that hop; it is useful to attach the MTR report when writing to support.
- Different routes from different countries — compare locations in the results: the problem may exist only in one region.