ICMP Ping
Ping is a plain ICMP Echo Request (three packets, five-second timeout) — not a TCP probe dressed up as a ping.
Access for monitoring checks
Add these to your firewall allow-list so our checks are not mistaken for unwanted traffic. The list is machine-readable, so a rule can be built by a script rather than by email.
81.163.20.249
Russia, Moscow ru-msk
Last changed: 2026-09-16
Last confirmed by measurement: 2026-10-10.
/en/monitoring-ips.txt
one address per line, comments start with #.
IPv6 is not used: all checks go out over IPv4.
Machines we run ourselves, one per country. These addresses are ours, so we publish them: a check arriving from one of them is ours.
38.180.26.199
Dubai, AE
own-ae
38.180.128.164
Sydney, AU
own-au
38.180.220.165
Sao Paulo, BR
own-br
149.33.52.190
Frankfurt am Main, DE
own-de
38.180.148.240
Tokyo, JP
own-jp
38.180.74.244
Singapore, SG
own-sg
38.180.138.136
New York, US
own-us
Our own nodes are listed above: the Moscow server and the regional probes we run ourselves. Where we have no probe of our own, checks fall back to check-host.net nodes — that pool does not belong to us, it changes between checks, and we cannot promise it will stay the same tomorrow.
No fixed IP list
Publishing those addresses would mean promising someone else’s infrastructure. An allow-list that works most of the time is worse than none: it looks like it works.
ru-spb
Russia (St. Petersburg)
check-host.net
eu-de
Germany
check-host.net
us-east
USA (East)
check-host.net
ap-sg
Singapore (Asia)
check-host.net
eu-nl
Netherlands
check-host.net
Ping is a plain ICMP Echo Request (three packets, five-second timeout) — not a TCP probe dressed up as a ping.
We ping the address the name resolved to, not the name itself. This is deliberate — it closes DNS rebinding — and it has a side effect worth knowing: with several A-records, one check hits one address, not all of them.
HTTP checks arrive with the User-Agent of our monitor, so they are easy to tell apart in your logs.