---
title: Thresholds by region for every network-distance alert
description: A DNS answer that takes 9 ms from Oceania takes 53 ms from North America, and a TCP connect of 2 ms in one region takes 137 ms in another. One threshold for…
canonical: https://watchfor.io/changelog/2026-11-19-thresholds-by-region-every-type
---

[All updates](/changelog/page/2#2026-11-19-thresholds-by-region-every-type) ⚡ Improved Alerting API Nov 19, 2026 · v 13.4.7

# Thresholds by region for every network-distance alert

A DNS answer that takes 9 ms from Oceania takes 53 ms from North America, and a TCP connect of 2 ms in one region takes 137 ms in another. One threshold for every location either alerted on distance or sat so high it missed a real slowdown near your users. Every latency, packet loss and jitter alert now works like the game server cards: a Now line per region, a table with each region judged on its own, and Suggest from data fitted to a week of real checks.

## Which alerts

- HTTP and API response time

- Ping (ICMP) latency and packet loss, UDP first byte, packet loss and jitter, game server jitter, network path (MTR) destination round trip and loss

- TCP connect and TLS handshake, DNS lookup, CDN time to first byte, WebSocket handshake, ping and total time, MCP handshake, NTP network delay

- SMTP , IMAP and POP3 connect, TLS handshake and login (IMAP also mailbox open), FTP / SFTP connect and login

Offline and every other rule stay judged from every region.

## How it works

- Now: Europe ~18 ms · Africa ~140 ms under the sliders, and a notice when one region is over a threshold while the others aren't.

- Different thresholds by region turns the card into a table: each region judged on its own, its own value or alerts off, confirmed only from its own locations. Each card has its own range (DNS from 5 ms, HTTP from 50 ms to 30 s).

- Suggest from data uses each region's history up to 7 days. Web and login timings: warning = max(p95 × 1.5, p95 + 30 ms, p99 × 1.2), critical = max(p95 × 2.5, p95 + 80 ms, p99 × 2). Network timings (DNS, TCP, ping, mail connect) use +15 / +40 ms. Jitter: max(p95 × 2, p95 + 5) / max(p95 × 3, p95 + 15). Packet loss moves one lost packet at a time.

- Switching one of these cards on creates a warning and a critical rule; new critical rules confirm after 3 checks in a row, like warning. Rules you already have keep their levels and counts.

## Fixed on the way

- The collapsed game server card listed regions the monitor isn't checked from as "off" in its summary. It now names only the regions it checks from.

- On DNS and TCP monitors the built-in Slow DNS Lookup / Slow TCP Connect rule now belongs to the card, so switching a region off really stops it alerting there.

- A region added with a new location stays off on one-level cards, also after you switch the other level on.

- Game server latency suggestions use the web formula above, so they come out a little higher than before (Africa at 173 ms: 260 / 440 ms instead of 230 / 320 ms).

## API and MCP

regions is accepted on every rule whose metric is marked region_scoped: true in the monitor-type catalog. GET /v1/monitors/{id}/suggested-thresholds?metric=… and the MCP tool get_suggested_thresholds take any of them. A network-distance rule created without failureThreshold confirms after 3 checks. SDKs 0.25.0. Details: [alert rules by region](/docs/alerting#alert-rules-by-region).

Share this update

---

Canonical page: https://watchfor.io/changelog/2026-11-19-thresholds-by-region-every-type · Site guide: https://watchfor.io/llms.txt
