Skip to main content
Updated October 19, 2026 Raff Apps tells you what went wrong in plain words, on the app page and on the Deploys tab. Find the message below. If your problem is not listed, contact support.

​
Deploys

What’s happening: the app started, but nothing answers on the port Raff sends traffic to, so the deploy failed.Why: the app listens on a different port.Fix:
  1. Make the app listen on the port in the PORT environment variable (8080 unless you set another). In Node.js, for example: app.listen(process.env.PORT || 8080).
  2. Or set the port your app really uses in Settings, Runtime, HTTP port. Saving it starts a new deploy.
What’s happening: the full message is Raff could not pull the image "<image>". Check the image name and tag, and that the image is public.Why: the image name or tag is wrong, or the image is private. Raff Apps pulls public images only.Fix: check the reference with docker pull <image> on a computer that is not signed in to the registry. Make the image public, then create the app again with the right reference, or deploy the corrected image through Create deployment.
What’s happening: the message shows the last lines of the build log. A line such as The command exited with code 1. means a step of your build failed.Fix:
  1. Open the Deploys tab and press Log on the failed deploy for the whole log.
  2. Fix the error in your code or Dockerfile and push again, or press Retry build.
  3. If the build needs settings, such as NEXT_PUBLIC_* or VITE_* values, give those variables the build or both scope. See How Raff builds and runs your code.
If the message says the log is no longer available, deploy again and follow the log on the Deploys tab.
What’s happening: for example, Raff could not find a Dockerfile at "Dockerfile". Check the Dockerfile path in the app's settings.Why: the Dockerfile path points at a file that is not in the repository, on that branch, in that folder. The path is relative to the app’s root directory.Fix: in Settings, Repository, correct Dockerfile path and Root directory, then redeploy.
What’s happening: builds stop after 15 minutes.Fix: make the build faster: install only what the app needs, and leave tests and other slow steps to your CI. Automatic builds (without a Dockerfile) reuse a cache from the previous build. If the build cannot be shortened, contact support.
What’s happening: Raff could not fetch the code it had staged for this deploy.Fix: press Retry build or push again. Contact support if it keeps happening.
What’s happening: the app starts, then exits. Raff starts it again, and it exits again.Why: an error at start, such as a missing environment variable, a wrong start command or a database it cannot reach.Fix:
  1. Open the Logs tab: the last lines before each exit show the error.
  2. Check the start command in Settings, Runtime, and the variables on the Environment tab.
  3. Fix it and redeploy.
What’s happening: the full message names the memory of the app’s size, for example Your app ran out of memory (512 MiB on its size) and was stopped.Fix: choose a larger size in Settings, Instance size, or make the app use less memory. See App sizes, scaling and sleeping.
What’s happening: for example, Your app is running but does not answer the health check at "/healthz" on port 8080. Check the health check path and the port.Why: the path does not exist, it returns an error status, or the app listens on another port.Fix: open the path on your app locally and check it returns a status from 200 to 399. Then correct Health check path or HTTP port in Settings, Runtime. Saving starts a new deploy.
What’s happening: the app did not answer in time, and nothing else explains it.Fix: check the Logs tab for what the app was doing. If it needs longer to set up after its port opens, add a health check path: an app with one has 10 minutes. If the logs show nothing, contact support.
What’s happening: an app that was live stopped answering.Fix: check the Logs and Metrics tabs for errors or memory at the size’s limit. Redeploy, or choose a larger size. If the logs show nothing, contact support.
What’s happening: ports 8008, 8012, 8013, 8022, 9090 and 9091 are reserved on web apps and on private apps that scale to zero.Fix: make your app listen on another port, such as 8080, and set the app’s port to match.
What’s happening: some templates need more memory: Metabase and LibreTranslate need at least Standard (1 vCPU · 2 GB), Stirling-PDF at least Starter (0.5 vCPU · 1 GB). The create page turns smaller sizes off and marks them Too small for the template.Why: on a smaller size, set later in Settings or through the API, the app runs out of memory, restarts or never turns ready.Fix: in Settings, Instance size, choose at least the template’s smallest size. See Raff Apps templates.
What’s happening: apps that run a Docker image are not built, so a variable with the build scope has nowhere to go.Fix: give the variable the runtime scope.
What’s happening: you resumed an app that never went live, so there is nothing to start.Fix: press Retry build or Redeploy to deploy it.

​
Cron and job runs

What’s happening: the command of a cron or job app ended with an error. The message says how many times it was tried.Fix: find the run on the Runs tab and read the Logs tab around its time. Fix the command or its variables, then press Run now.
Fix: choose a larger size in Settings, Instance size, or make the command use less memory, for example by working in smaller batches.

​
Names and limits

What’s happening: app names are unique within a project.Fix: choose another name, or create the app in another project. The create page suggests a free name, such as n8n-2.
What’s happening: an account can run 20 instances across its web, private and worker apps. Paused, failed and not yet deployed apps count too, because resuming starts them again.Fix: scale down or delete an app you do not need, or contact support to raise the limit. Scaling down is always allowed.
What’s happening: one app can run 6 instances, fixed or as its autoscaling maximum.Fix: use a larger size for more capacity per instance, or contact support.
What’s happening: other apps reach a private app on its port, so Raff needs it at create.Fix: enter the port your app listens on, for example 8080.
What’s happening: on monthly billing, the size is part of the app’s plan and cannot be changed from the dashboard yet.Fix: contact support to change the plan.

​
Spend cap

What’s happening: your account’s app usage reached its monthly spend cap, and the cap is set to Pause my apps. Pay-as-you-go apps are paused with their data kept. Monthly apps within their plan keep running. Until the cap allows it, Raff refuses to resume, create, deploy or scale up these apps.Fix: on the Apps page, press Manage cap and raise the cap, turn it off, or choose Keep apps running and email me. The paused apps start again within seconds. Otherwise they start again next month. See Set a spend cap.

​
Databases

What’s happening: the database refuses connections without TLS, and Node.js pg does not use TLS unless you ask for it.Fix: pass the database’s CA when you create the pool: new Pool({ connectionString: process.env.DATABASE_URL, ssl: { ca: process.env.DATABASE_CA_CERT } }). Most other clients, such as psql, psycopg, pgx, Prisma and JDBC, use TLS by themselves. See How apps connect to databases.
What’s happening: links made before October 15, 2026 connect by IP address and have no DATABASE_CA_CERT, so the app cannot check the certificate.Fix: on the Bindings tab, rotate the link’s credentials. The app redeploys with the new password, the host name and DATABASE_CA_CERT.
What’s happening: an app created with a database waits for it to run, up to 30 minutes. If the database is not running by then, the app deploys without it.Fix: when the database runs, attach it on the app’s Bindings tab. Attaching redeploys the app with the DATABASE_* variables.

​
Sleeping, domains and network

What’s happening: the app has Scale to zero on. It sleeps about 15 minutes after its last request, and the next request waits a few seconds while an instance starts.Fix: if every request must be fast, turn off Scale to zero in Settings, Scaling. On pay-as-you-go, the app is then billed all the time it runs.
What’s happening: Raff cannot see the right DNS records yet. The message under the domain says which one:
  • We could not find the TXT record _raff-verify.<domain>: add the TXT record with the verification value shown.
  • The TXT record ... does not match: copy the value into the record exactly.
  • This domain does not resolve yet: add the CNAME record to the target shown.
  • This domain is proxied through Cloudflare (orange cloud): set the record to DNS only (grey cloud).
  • This domain points to ..., not ...: change the CNAME, or the A record, to the target shown.
Fix: correct the records at your DNS provider, then press Check now. For a root domain, use an A record to the target’s IP or your provider’s ALIAS record. After 24 hours without working DNS the domain turns to Error; fixing the records and pressing Check now retries it. See Add a custom domain.
What’s happening: DNS was verified, but the certificate could not be issued. Usually the record changed or became proxied after verification.Fix: check that the CNAME still points to the target shown and is not proxied, then retry with Check now.
What’s happening: a domain can serve one app at a time.Fix: remove the domain from the other app first, or use another subdomain. If the other app is not yours, contact support.
What’s happening: outbound TCP port 25 is blocked for every app.Fix: send through your email provider’s submission port (587 or 465) or its HTTP API.

​
GitHub

What’s happening: the Raff GitHub app was uninstalled from the GitHub account, or its access was removed. Pushes no longer deploy; the app keeps running its last deploy. The app page shows a banner with Reconnect GitHub.Fix: press Reconnect GitHub and sign in to GitHub. Raff then deploys the latest commit on the branch if it is not already running. See Reconnect GitHub.
What’s happening: an owner of the GitHub account suspended the Raff GitHub app, so pushes do not deploy.Fix: press Open GitHub settings on the banner and unsuspend the app there.
What’s happening: the repository was removed from the Raff GitHub app’s access, or it was renamed, moved or deleted.Fix: in GitHub, give the Raff GitHub app access to the repository, then deploy again. If you renamed or moved it, press Reconnect GitHub on the app page.
What’s happening: the branch does not exist, or Raff has no access to the repository.Fix: check the branch name in Settings, Repository, and that the Raff GitHub app still has access to the repository.
What’s happening: Raff builds repositories up to 100 MB.Fix: remove large files from the repository, or build the image in your own pipeline and deploy it as a public Docker image.

​
Still stuck?

Contact support@rafftechnologies.com with:
  • The app ID, from the dashboard URL
  • The exact message and the time it happened
  • What you have already tried
Last modified on October 19, 2026