Ping monitoring: read ICMP reachability, RTT, and packet loss
Use Ping monitoring to check a path with ICMP. Interpret round-trip time and sampled packet loss, pair results with HTTP checks, and set useful alerts.
A branch-office probe stops getting replies from the gateway, but the HTTPS check still passes. Start by checking whether those probes ran from the same network and reached the same address. Then check whether the gateway filters or rate-limits ICMP. That combination of results can narrow the investigation without declaring the service down.
Ping monitoring sends ICMP Echo Requests on a schedule and records the matching Echo Replies. It measures reachability as observed from a particular probe, together with round-trip time (RTT) and sampled echo loss. An HTTP check asks whether an application responds as expected. The two signals belong on the same timeline when you're investigating a network problem. Uptime monitoring
Measure the path from the probe
IPv4 ping uses ICMP Echo Request and Echo Reply, and IPv6 uses ICMPv6 equivalents. If you enter a hostname, the probe must resolve it before sending requests. Record the address and IP version used for each test. A probe inside the branch and one on a public network can reach different addresses or take different paths even when they start with the same name. ICMPv4 specification · ICMPv6 specification
To check your local IPv4 ping command on Linux with iputils or on macOS, use the loopback address:
ping -c 3 127.0.0.1
On Windows, use ping -n 3 127.0.0.1. Timeout flags differ too. Linux iputils uses seconds for -W and macOS uses milliseconds for that option. Loopback only verifies local behavior. To reproduce a monitoring result, run a permitted test from the affected network and compare the destination and IP version. Linux ping manual · Windows ping documentation
Read RTT and loss together
RTT starts when the probe sends a request and ends when it receives the matching reply. The result includes the outward path, processing at the target, and the return path. A single measurement cannot tell which leg got slower. Compare it with a baseline for the same probe, address, and IP version.
For an illustrative test, send five requests and wait for their replies. If four matching replies arrive in time, the observed echo loss is one divided by five, or 20%. Calculate RTT from the four valid replies. Keep a missed or unreported run separate from a run that sent requests and received none. With a small packet count, one missed reply moves the percentage sharply. Read several runs before you attribute the change to a network fault.
An explicit unreachable message and a silent timeout are different observations. A timeout can come from a down target, a route or policy problem, a firewall, or an ICMP reply rate limit. A device can also treat echo traffic differently from application traffic. The loss figure applies to these probes and their waiting time. IPv4 router guidance on rate limiting
Pair Ping with a service check
| What you check | What a successful result tells you |
|---|---|
| ICMP Echo | The probe received an Echo Reply from the target within its waiting time |
| TCP connection | The probe could establish a connection to the target port |
| HTTPS request with a response assertion | The request completed and met the configured checks, such as status or response content |
For the branch gateway, compare checks from the affected branch before using a healthy public probe to clear the incident. If Ping works and HTTPS fails from the same location, investigate the TCP connection, TLS handshake, and application response in order. If both fail, check shared dependencies such as name resolution or routing. A certificate or response assertion belongs in the application check, and domain registration expiry is a separate signal. API monitoring · Domain expiration monitoring
Configure the test and the alert separately
In Guance, create an ICMP synthetic test with the target hostname or address, a packet count from 1 to 10 per test, availability conditions, the relevant nodes, and a schedule. The setup page's Test action runs independently of the selected nodes. Inspect the scheduled results from the locations that actually matter. Guance documents an RTT display of 0 when there is no valid RTT, so exclude that case from normal latency statistics.
The task schedule determines when packets go out. The synthetic test monitor evaluates reported results over a detection window at its own frequency. Set the window at least as long as that frequency and cover the actual reporting period. Choose thresholds against a baseline from each path and the impact on the service. If transient loss is expected, use a consecutive trigger condition for sustained degradation and decide separately how the monitor handles missing data. The alert should identify the affected location so the responder can reproduce the path.
Contact us
Join the community
to join the community
Try Guance
Start online and pay only for what you use.
Get startedChoose a Guance plan
