Skip to content

Ping monitors

A Ping monitor sends an ICMP echo request — the same probe as the classic ping command — to a hostname or IP address, and counts the check as up if an echo reply comes back within the 10-second timeout. (For the protocol details, see how ping checks work.) It's the lightest check UpAll offers and the natural choice for servers, switches, routers, firewalls, and internet links — anything whose core question is "is this reachable over the network?"

What a successful ping proves

An echo reply establishes three facts: the device is powered on with a working network stack, the path between UpAll and the device carries traffic in both directions, and the round trip took the number of milliseconds recorded as the response time.

That covers most hard failures — power loss, a crashed OS, an unplugged uplink, and a dead ISP circuit all stop ping immediately. What a reply does not prove is that any particular service on the device works: a server can answer ping flawlessly while its database or mail daemon is down. If a service is what you actually care about, monitor the service — an HTTP monitor for a website, a TCP port monitor for anything else. For critical hosts, run both: when the service check fails but ping still answers, you know instantly that the fault is in the software, not the machine or the network.

When a firewall blocks ICMP

Plenty of networks drop ICMP at the perimeter, and some deprioritize it under load. That has two consequences for monitoring:

  • A blocked target looks down when it isn't. If Test connection fails on a host you know is up, or the monitor shows permanent failures while the system works, ICMP filtering is the usual suspect. The fix is to switch to a TCP port monitor aimed at a port you know is open — it measures reachability just as well and passes through firewalls that permit the service.
  • Deprioritized ICMP distorts latency. A device that answers ping lazily under load can show alarming response times while its real services stay fast. Judge latency trends with that in mind.

The degraded threshold for ping

Ping round-trip time is pure network time — no application processing — so healthy values are small: a few milliseconds on a LAN, a few dozen across the country. That's why the degraded threshold is capped at 1999 ms for Ping monitors (HTTP and TCP allow up to 9999): a ping that takes anywhere near 2 seconds already signals a saturated link or a struggling device. Ping's recommended Service class starts at 100 ms for exactly this reason — for LAN gear, that already surfaces congestion long before users complain. Raise it toward the cap only for links where higher latency is expected, like a distant branch office reached over the public internet.

As with every check type, one slow or lost ping doesn't open an incident — that takes the monitor's configured streak of consecutive failures; see Check intervals and thresholds.