Skip to content
shadcn.io

shadcn.io is not affiliated with official shadcn/ui

System Alerts by Severity Notification Block for React

An infrastructure monitor's alert panel: a count of open alerts by severity, and an accordion of three systems with their facts, an Acknowledge and a one-hour Mute that says when it ends.

Framework
UI library

Preview

Preview width

Installation

Part of Pro
Pro unlocks the code and one-line install of every Pro button and text effect, plus the MCP for your AI editor.
Have Pro already? Sign in to open it.Opens here as soon as you're Pro.

Overview

Group what is firing by system and say how bad it is, with this system alert panel for React with Base UI. It is Rackbell, an infrastructure monitor, in a hairline frame with a plus mark on each corner. The first band counts what is still open in three cells: Critical (heavier, with its own glyph), Warning and Quiet, which is what someone acknowledged or muted. Under it an Accordion holds three systems, the worst first: each header has the severity glyph, the name, the severity as a Badge word and the host with its state, and the open one shows a Data List (host, region, since, the metric that tripped) with an Acknowledge and a Mute 1 hour Button. Acknowledging stamps a time and muting says "until 10:30 UTC", an hour after the page's fixed now. One panel is open at a time and every panel is the same height. The lines come from Guides, and nothing is timed. For monitoring consoles, status boards and incident tools.

FAQ

How do I fill it with my systems?

Edit SYSTEMS: each has an id, a level of critical or warning, a name, a host, a region, an ISO since, the metric that tripped and a sentence for what. Keep them worst first. AS_OF is the page's fixed now and MUTE_MINUTES is the length of a mute, so the "until" time is worked out from them and reads the same everywhere. The three counts are derived from the systems, the acknowledgements and the mutes.

What do Acknowledge and Mute do?

They mark the system in this tab: Acknowledge stamps a demo time (a minute after now for the first) and its button becomes Undo acknowledge, Mute 1 hour says when the mute ends and becomes Unmute, and the three counts move. Nothing is sent or saved, and a reload brings everything back. In your app call your monitor from acknowledge and mute, and take the time from its answer.

Why does nothing time out?

A persistent alert should stay until someone acts on it, so there is no ticking clock and a mute is a typed length, not a timer. When a mute ends in your app your server would bring the alert back; here the page simply states when.

Does opening another system move the page?

No. Every panel has the same structure and the same minimum height, one is always open, and the headers stay where they are. Tab moves between the headers; arrow keys do not.

How is severity shown without colour?

By word, glyph and weight: Critical has a solid badge, a bold figure in the summary and a semibold name; Warning has an outline badge and a medium figure; Quiet is regular and muted. Only the critical glyph takes the destructive status colour.

Is it accessible?

Each header is a heading with a button inside, the buttons name their system ("Acknowledge: Database"), and a polite status line gives the new counts after each action ("1 critical, 1 warning, 1 quiet"). The facts are a description list, so every value keeps its label.

  • Pro

    An on-call paging panel with three urgent alerts that demand an answer: severity as a word, a glyph and a weight, a fixed escalation time, and an Acknowledge that stamps the time and takes the escalation line away.

  • Pro

    An outage notice with the facts of the incident in a quiet column (status, duration, start, regions as chips), an ordered list of updates in a frame, and a Subscribe to updates with Email and Push checkboxes that answers in the demo.

  • Pro

    An error tracker's grouped issues in a table: severity as a word and a glyph, event counts and dates, a status menu that updates the word and moves the issue, and an issue that opens to its stack in a numbered code window with a Copy for the event id.

  • Pro

    Alert Banner Notification

    Incidents and alerts

    An operations console's banner bar over a sketch of its app: four notices queue by urgency and show one at a time, with Previous, Next, a link, Dismiss and Snooze all for an hour.

  • Pro

    One failed request from an API gateway: a 502 with the endpoint, upstream, time and request id (with a Copy), tabs for the request, the response and the headers in numbered code windows, and a bounded simulated Retry.

  • Pro

    A metrics service's fired alert rule you can edit: a slider moves the threshold over the last twelve CPU readings, the bars that cross it are marked and counted, and the rule rewrites itself before you save it.