Crescendo Lab Docs
English
Sign up

Versions and going live

Every upload creates a new version. Versions are never overwritten, and going live always puts the newest version live.

Last updated

Build stages#

After an upload, a version moves through these stages. Its row on the Deploy tab shows the current one.

StageWhat the platform is doing
QueuedWaiting for a build machine. The row shows how many builds are ahead.
BuildingBuilding the web page or the container image.
ScanningChecking for secrets, personal data and scripts loaded from elsewhere. A container image is also scanned for vulnerabilities and secrets.
ReviewingCollecting the scan results.
Awaiting confirmationA check set to "asks first" found something. After reading the findings, you can publish anyway.
SucceededThe version can go live.
CancelledA member cancelled it while it was queued, so it was not built. The row names who and when.
FailedThe content was refused or the platform failed. See Troubleshoot failed deployments.

Each check in a version's results says how it is handled: "must be fixed before going live", "asks first" or "not run".

Go live#

When your workspace has only the production environment, every new version goes live on its own once it builds, unless a newer version is already live. Visitors see it within a minute.

Going live always puts the newest version live. When the newest version isn't live yet, for example because it was only built or was uploaded to another environment, select Go live on the newest-deploy notice of the Deploy tab, or on its row. Only the newest version can go live. To run earlier code again, deploy it again as a new version (see Go back to earlier code).

When the workspace has more than one environment, a version goes live only in the environment you upload it to (see Environments).

If a version has a problem, fix it and deploy again. To try a version before visitors see it, first deploy it to a separate environment, such as staging, and go live in production once it works.

Every version keeps its own preview address, whichever version is live.

Internal calls move too

Other projects that call this project's internal address reach the live version. Going live changes where those calls land.

Cancel a queued build#

While a version is still Queued, select Cancel build on the newest-deploy notice of the Deploy tab, or on the version's own page. It is cancelled at once, without a confirmation: it is not built, it doesn't use build minutes, and what is live doesn't change in any environment. Anyone who can deploy the project can cancel.

A cancel can be taken back. Within about a day of the upload, select Build again where the cancelled version is shown, and it is queued again without uploading. After that, the uploaded files are gone; deploy again instead.

Once a build machine has started the build, it can't be cancelled. It finishes as usual; if you don't want that version, deploy a fixed one after it finishes.

A version built from a GitHub push stays cancelled: GitHub sending the same push again doesn't bring it back. Deploy latest commit or a new push builds that commit again.

Go back to earlier code#

A version that finishes building after a newer one is already live does not go live by itself. This happens when two pushes, or two uploads that go live in the same environment, are close together and the later one builds faster. The newer version keeps serving. The earlier one stays in the list as an ordinary earlier version, with "Not put live: version #N was newer". Nothing failed and you don't need to do anything.

To run earlier code again, push it again (or upload it again). That makes a new version, newer than the live one, and it goes live like any other.

Environments#

When a project has more than one environment, every action applies to the environment you are on. The upload button reads "Upload and go live in ‹environment›", and going live affects only that environment. Each environment has its own settings and secrets.

Before a version goes live in an environment, the platform compares the environment's variable and secret names and tells you about any that are missing.

Before a new environment takes traffic

A new environment has an address of its own, so anything outside Agent Lab that knows your production address has to learn the new one. Check these before you hand it over:

  • Values it came without. An environment copied with secret names only lists each missing secret as an empty row in its Variables, naming the environment it was copied from, and the project overview's status stays Needs attention until they are set. Meanwhile the environment already runs the version its source serves.
  • OAuth callback URL. If the project signs people in with Google, GitHub, LINE or another OAuth provider, add the new environment's address with your callback path to that provider's allowed redirect (callback) URLs. Otherwise sign-in fails with a redirect mismatch.
  • Sign-in domain. If your sign-in service keeps a list of allowed domains, such as Firebase Authentication's authorized domains or an allowed-origins list, add the new address there too.
  • Addresses in values. Copied values that pointed at the source environment's Agent Lab address or database are rewritten for the new environment. An address of anything outside Agent Lab is copied as it is.