ICMP monitors

What it checks

Sends ICMP echo requests to a host and reports the median round-trip time. The check fails when packet loss exceeds your threshold, or when nothing comes back at all.

Median rather than mean, deliberately: one retransmit on a four-packet sample moves a mean noticeably, and the number an operator glances at should reflect the typical case.

Fields

A threshold of 0 means any lost packet fails the check. On a link you know is lossy, 25 or 50 reports genuine degradation without paging you for one dropped echo.

Create it with the API

curl -X POST https://api.whatping.com/v1/monitors \
  -H "Authorization: Bearer $KEY" \
  -H "content-type: application/json" \
  -d '{
    "name": "gateway",
    "type": "icmp",
    "host": "gw.example.com",
    "packet_count": 4,
    "loss_threshold_pct": 25,
    "interval_sec": 60
  }'

Field names are snake_case and an unknown one is a 422 naming the field, never a silent drop. Full reference: API.

What it tells you, and what it does not

ICMP proves a host is reachable and how far away it is. It says nothing about whether the service on that host works — a machine with a full disk and a dead application answers pings perfectly.

Use it for network reachability: a gateway, a router, a VPN endpoint, or a host you want latency history for. For anything serving traffic, an HTTP or TCP monitor is the stronger signal.

Many networks also deprioritise or drop ICMP entirely. A host that fails an ICMP check but passes an HTTP check is usually a firewall policy, not an outage.

Failure messages

100% packet loss (4 sent, 0 received)
50% packet loss exceeds 25% threshold
host did not resolve

Running unprivileged

Worth stating, because it is the usual reason a monitoring agent ends up running as root: this one does not. ICMP here uses a SOCK_DGRAM socket, which Linux permits for process groups inside net.ipv4.ping_group_range, and the deploy grants that to the worker’s group alone — not to every group on the host, and without CAP_NET_RAW on the binary.

If ICMP monitors report icmp socket unavailable, that sysctl is the thing to check.

Worked example

Host:                   gateway.example.com
Echo requests:          4
Packet-loss threshold:  25%
Interval:               60 s
Failures before down:   2

Related

• TCP monitors
• Limits