Urgent.News

What's breaking now, across hundreds of outlets.

Tech

Building Sentinel: Uptime Monitoring with Signed Reports and Quorum

The problem with a single uptime checker Most uptime monitors are one process making one HTTP request from one location. That is enough to catch most outages, but it has two blind spots: the checker itself can go down, and regional network problems look like global outages. If my checker is in us-east and an ISP there has a routing issue, I could page my team at 3 AM for a problem that only a…

Most uptime monitors are a single process that makes a single HTTP request from one location. This setup is sufficient to catch most outages, but it has two major blind spots. The first blind spot is that the checker itself can go down, and the second is that regional network problems may be mistaken for global outages. To address these issues, I developed Sentinel to explore whether a small network of independent validator nodes, each checking the same target from a different region and signing its report, could reach a majority agreement and avoid false alarms caused by regional noise.

Here's how Sentinel works. Each validator node, such as node-us-east, node-eu-west, and node-ap-south, fetches its assignments from a hub and performs an HTTP health check on the target. The hub then receives a signed report, which includes the current window ID, a random nonce, and the result of the health check. The signed report is validated, aggregated within a configurable time window, and then passed through a consensus engine that determines a verdict based on a quorum and majority agreement.

The key design decisions in Sentinel include using quorum instead of relying on a single node, signatures to prove the identity of the report sender, and an "inconclusive" verdict when not enough reports are received within the deadline. Quorum ensures that at least a majority of reports must agree for a verdict to be considered valid, while signatures confirm the authenticity of the report sender. An inconclusive verdict is provided when there are not enough reports to reach a quorum, preventing premature alerts.

Rewritten by urgent.news; may contain errors — check the original.

Source: dev.to

Read the original at dev.to →

Report a problem with this story

More in Tech

Fix: iOS Codex Shows 'No Tasks Yet' Bug

Originally published on DebugRadar . Quick Fix Summary Problem: Codex on the ChatGPT iOS app shows "No tasks yet" / "Start a task to see it here" while the host still shows connected (green).

  • iOS Codex app shows "No tasks yet" bug while sessions are connected
  • Issue reported November 11, 2026, persists as of November 15, 2026
  • Resolved by using search function or notification links in app

Fix: Codex CLI zstd Compression 400 Error

Originally published on DebugRadar . Fix: Codex CLI zstd Compression 400 Error Quick Fix Summary Problem: Codex CLI requests fail with a bare 400 {"detail":"Bad Request"} on…

A small test fixture for Chinese SRT subtitle imports

A subtitle import problem is easier to isolate with a tiny fixture than with a full video transcript. This example separates three checks: whether the editor can decode Chinese characters, parse cue timing, and keep line breaks. It is a proposed diagnostic fixture, not a report of tests across editing applications.

More from Sunday 22 November →