The first hour after a deploy

Updated 1 October 2026 · BugsRadar team

The deploy went green. Now the real test: an hour of refreshing logs and wondering whether the quiet means it works or means nobody has hit the new code yet. The hour goes better when the chat tells you three things: that the deploy finished, which version an error comes from, and whether an error is new or came back.

The deploy is a message

Add a step to the pipeline that sends an Information message when the deploy succeeds, with the commit hash and the environment: see Report finished deploys. Each deploy is a message of its own, because the hash differs, so the chat shows "deployed a1b2c3 to production at 14:02" and then whatever happened next. A deploy that fails sends its own message too, with the link to the run.

The version is in every error

Every message carries the application version next to the environment and the host: Production · web-1 · v1.4.3. The .NET package takes it from the entry assembly by default; in Node.js and Python you pass appVersion or app_version. An error from the old build and the same error from the new one are told apart at a glance, and "did my deploy cause this" is answered by reading one line.

A regression looks new

When an error stops happening and a repeat window passes without it, BugsRadar closes it. If the deploy brings it back, it arrives as a new first message with the stack trace, not as a count edited into last week's thread. The errors that keep going through the deploy keep counting in their old message, so the chat separates "still broken" from "broken again" for you. See What you receive.

Staging first

Deploy to staging, watch its channel for ten minutes, then production. Set the environment in the package configuration; on the Pro plan a rule gives staging its own channel so that a staging error never reads like a production one. The same error in the two environments is counted separately either way.

The hour, in messages

14:02 deployed. 14:09 a new error, v1.4.3, NullReferenceException in the new checkout step, ×1. 14:11 ×14. 14:25 rolled back, deployed a0b1c2. 14:26 nothing. 15:02 still nothing, and you know that because the chat would have said otherwise.

Where to go next

The pipeline steps are in Alert when a CI/CD pipeline fails. Production, staging, dev covers the channels per environment.