Ping monitor

The ping monitor sends ICMP echo requests to a host and records latency and packet loss. It's the cheapest, most generic reachability check available — use it for network-layer "is this box even up" alerts, not for service health.

Overview

Every ping check records:

• Average round-trip latency in milliseconds across the configured packet count.
• Packet loss percentage (received vs sent).
• Status: up if at least one reply came back within the timeout, otherwise down.

Internally, the agent shells out to the system ping binary (which has the right capabilities to send ICMP from a non-root user). The check runs ping -c <count> -W <timeout> and parses the summary line.

Configuration options

• Host — Hostname or IP address. DNS is resolved on every check.
• Interval — How often to run the check (default: 60 seconds).
• Packet count — Number of echo requests per check (default: 3).
• Per-packet timeout — Seconds to wait for each reply (default: 2s).
• Regions — See Multi-region checks.

When to use ping

Ping tells you whether a network host is responsive — nothing more. Reach for it when:

• You need a basic liveness signal for a router, switch, gateway, or VM where no HTTP service runs.
• You're correlating latency from multiple regions to spot routing issues.
• You're monitoring a host that's deliberately not running any L7 service.

For anything customer-facing (a website, an API, a database), pair ping with a service-level monitor. A successful ping doesn't mean your service is working — it just means the kernel is alive enough to reply.

When not to use ping

• Cloud load balancers and managed services often don't respond to ICMP. AWS ELBs, Cloudflare, and many CDN edges drop ICMP at the network edge. Use HTTP instead.
• ICMP is rate-limited or deprioritized on the public internet. Bursty packet loss or spiky latency on a ping monitor often reflects the upstream network's policing, not your service.
• Some networks block ICMP entirely. If your monitoring agent's region can't send ICMP (uncommon but possible), the check will fail consistently regardless of target health.

Example

You manage a fleet of edge routers. Create a ping monitor for each:

Type:     Ping
Host:     edge-router-iad-1.internal
Interval: 60s
Count:    3
Timeout:  2s

A down result means three consecutive ICMP echoes were dropped or didn't return within 2 seconds — usually a hard network failure.

Limitations

• ICMP only. No TCP, no UDP, no application-level signal.
• Latency is the average across count packets — outliers are smoothed out. If you need percentile latency, monitor the underlying service with HTTP or TCP instead.
• Some firewalls log ICMP at high volume. With the default 60-second interval and 3 packets per check, that's 4,320 ICMP packets per day per region — usually fine, but worth knowing.