Skip to main content
A deployment can fail during the build, at the provider, or after the new version starts. Use these steps to find where it failed before you change anything in production.
Start from the failed deploy and read its steps, logs and operations. If the provider’s answer is unclear, check the provider and settle the operation before retrying. If it is known, review the smallest fix, deploy it, and check the running app.Start from the failed deploy and read its steps, logs and operations. If the provider’s answer is unclear, check the provider and settle the operation before retrying. If it is known, review the smallest fix, deploy it, and check the running app.

Start from the deploy’s page. If Hoista never heard back from the provider, check the provider before you deploy again.

1

Open the failed deploy

In Desktop, open the deploy from Deploys. The page shows the steps it got through, why it stopped under What went wrong, and the build output. If the cause is a missing secret, paste it there and deploy again. For a Hoista deploy, the live site is unchanged.
The waitlist-v2 deploy page, failed at the Secrets step, asking for RESEND_API_KEYThe waitlist-v2 deploy page, failed at the Secrets step, asking for RESEND_API_KEY

A deploy that stopped at Secrets. It names the missing secret and has a box to paste it.

If a provider does not provide logs, empty output does not mean success. Check the provider directly.
2

Ask an agent to investigate

3

Check an unclear outcome

If Hoista did not get a response, the provider may still have accepted the request. Check the provider’s release state and settle the operation before you retry.
4

Verify the fix

Check the new deployment’s result and the running app. A successful build does not prove that sign-in, database access or your main user flows work.If you need to roll back, check what that provider can restore, then run hoist rollback DEPLOY_ID or roll back from the project’s Deploys tab. Rolling back does not reverse database migrations.

​
Keep evidence safe

Note the deploy and operation IDs when you ask for help. Read a diagnostics report from hoist diagnostics before you share it. Do not publish raw logs that contain personal data or credentials.

Plan and deploy

Deploy again once it is fixed.

Logs and metrics

Read the error logs.

Backups and outcomes

Settle an operation with no answer.