See every GitHub deploy right next to the monitors and servers it affects, so you can spot a bad release the moment it causes trouble. This is for anyone who ships code through GitHub and wants deploys lined up with uptime and performance data, without switching tabs to compare timestamps.
#Overview
- Link a GitHub repository to a monitor, server or tag, and its deploys show up automatically from then on.
- Deploys appear on the incident itself, as markers on charts, and in an answer to whether the deploy made things worse.
- Linking a repository never changes a monitor's status, uptime or incidents — it only adds context.
NoteYou need to connect GitHub using the InfraNest Automations app. A plain token connection does not receive deploys.
#Link a repository
- Go to Integrations and open GitHub.
- Go to Connections & settings. Each connected GitHub account has its own Deploys on your monitors section. In the account that owns the repository, click Link a repository. If something is already linked, click Manage to see existing links and add more.
- Under Repository, pick the repository you want.
- Under What counts as a deploy, choose one of: A GitHub deployment, A workflow finishing, or A published release. If you're not sure, pick A GitHub deployment — if none are found, you'll be offered A workflow finishing instead.
- For A GitHub deployment, choose under Which deploys whether all deploys count or only one environment.
- For A workflow finishing, pick the workflow and, optionally, a branch.
- Under Shown on, pick a monitor, a server or a tag.
- Click Link.
Each link then shows its status: last deploy 3 days ago (working), looking for earlier deploys… (still fetching), no deploys found on GitHub (GitHub has no record of a deploy), or no deploys received yet (set up, just no deploy yet).
#Link a repository from a monitor
- Open the monitor and go to its Settings tab.
- In the GitHub deploys card, click Link a repository.
- Pick the Repository and What counts as a deploy, then click Link.
Linking a tag covers everything with that tag, including things you tag later. You can link the same repository more than once — for example to a monitor and a server. To stop showing deploys somewhere, remove the link from the same list or from the monitor's GitHub deploys card.
#What you see
- On an incident: deploys from the hour before it started are listed in the incident's timeline, with a link to the deploy and the pull requests it included.
- Under the response-time chart: one sentence tells you if a deploy made a location slower or faster, with a Compare before and after button for details.
- On charts: each deploy is a dashed vertical line with a flag. Hover to see details, or click to open the deploy's own page.
#Open a deploy's page
Click a deploy marker, a deploy listed under a chart, or a deploy named as a likely cause of a problem. The page shows:
- Did it make things worse? — the verdict for the hour after the deploy.
- What changed after this deploy — CPU, memory, restarts, response times and failed checks, compared with before the deploy.
- In this deploy — the pull requests it put live, each marked Likely, Possible or Unrelated.
- Where it runs — which servers the deploy reached, based on restarts seen by the agent, unless your pipeline reports versions directly.
Click Open a GitHub issue with this evidence to share what you found; you can edit the title and description first.
#Tell InfraNest exactly which version each server runs
Restarts are a best guess. To make it certain, have your pipeline report each server's version with your deploy token:
curl -X POST https://api.infranest.io/api/webhook/deploy/version \
-H "Authorization: Bearer $INFRANEST_DEPLOY_TOKEN" \
-H "Content-Type: application/json" \
-d '{"hosts": ["web-01", "web-02"], "version": "'"$GITHUB_SHA"'", "repository": "acme/shop"}'
#Mark a change as expected
If a change after a deploy was intentional — more memory for a new cache, say — click This is expected. The deploy then stops being flagged as the likely cause of that change. Click Undo to take it back.
#Redeploy the previous version
When a deploy broke something, you can roll back from InfraNest. It's off by default, since it changes production outside your pipeline.
- On the GitHub integration page, open Manage under Deploys on your monitors, in the account that owns the repository.
- Under Redeploy from InfraNest, switch it on for the link.
- On the deploy's page, click Redeploy abc1234, then confirm in the window that appears.
If it isn't offered, the page explains why and links to doing it directly on GitHub.
#Did a deploy make things worse?
After a production deploy, InfraNest checks the monitors it reaches more often for an hour, then compares that hour with the hour before the deploy, looking at response time and failed checks from every location.
You'll see one of these labels: Worse, Better, No change, Not enough data, or Checking.
Where you see it:
- On the deploy's page, with the full numbers.
- On the monitor's chart, as an orange marker for a deploy that made things worse.
- On Home, as an item like "Checkout got worse after a deploy", with a See the deploy link.
TipOnly production deploys get this extra checking, and InfraNest decides what counts as production from what GitHub tells it. If that's wrong for a repository, open Manage under Deploys on your monitors and set Which deploys are production to Always production or Never production.
These extra checks never run faster than your plan allows, and deploying several times within the hour won't multiply them — the earlier deploy's comparison window is simply cut short. They're ordinary checks: they don't affect your uptime, alerts or incidents.
#Good to know
- Deploys appear as soon as GitHub reports them. Earlier deploys are fetched automatically when you link the repository.
- Deploys are kept for as long as your plan keeps monitoring history.
- With PagerDuty connected, a linked deploy also appears under Recent changes on the PagerDuty incident.
- Only finished deploys show up — successful, failed or errored. Deploys still in progress don't.
#Troubleshooting
- The link says "no deploys found on GitHub". Your deploy job doesn't tell GitHub it happened. Either give it an environment (for example
environment: production), or remove the link and link again using A workflow finishing with your deploy workflow. - No deploys appear, and the link says "no deploys received yet". Check that GitHub is connected using the InfraNest Automations app, not a plain token.
- Deploys appear for the wrong environment. Remove the link and link the repository again, choosing the right environment under Which deploys.
Was this article helpful?
Thanks for your feedback!
Thanks, this goes straight to the docs team.