New Claude explains failed deploys

Your own cloud,
for every git push.

DeployPro builds, health-checks and ships your apps to a server you own. Previews for every branch, instant rollback, domains with HTTPS, logs and metrics, without the per-seat bill.

app.deploypro.us/projects/blog
B
Blogblog.acme.com · acme/blog · main
Healthy
OverviewDeploymentsRuntimeConfigurationDomains
Production#48 · a1b2c3d · “Add pricing page”
  1. Source
  2. Build
  3. Deploy
  4. Production
CPU18%
Memory312 MB
Network in4.2 KB/s

In production on DeployPro today

  • DispatchIts API, behind a custom domain
  • CineForgeFilm-rendering workers with Redis
  • BalanceVidVideo playout and streaming
  • deploypro.usThis page

Detected automatically, or bring your own Dockerfile

  • Next.js
  • Vite
  • Nuxt
  • Astro
  • SvelteKit
  • Remix
  • Node
  • Python
  • Go
  • Static
  • Dockerfile
  • Any image

Anatomy of a deploy

Four steps. Production only moves
when the new release is healthy.

  1. 01 · Source

    You push.

    The GitHub App sees the commit, puts a pending check on it and queues a build. A pull request gets its own preview.

    push · main · a1b2c3d
  2. 02 · Build

    It builds.

    Your Dockerfile, or one written for the framework it detects. Secrets are mounted, never baked into a layer.

    next build · 1m 52s
  3. 03 · Deploy

    It proves itself.

    The new container starts beside the old one and must answer its health check. If it doesn’t, nothing changes.

    healthy after 1.9s · 200 /
  4. 04 · Production

    Traffic moves.

    The router switches in one step: no dropped requests. The previous release stays ready for a one-click rollback.

    live · blog.acme.com

Collaborate

Every branch gets a live preview.

Open a pull request and DeployPro builds it at its own address, puts the link on the PR and a ✓ on the commit. Review the real thing, not a screenshot. Merge, and production follows.

  • A preview address per deployment
  • Commit statuses and a PR comment, edited on every push
  • Production untouched until the branch is merged

Operate

See everything. Undo anything.

Live runtime logs

Follow the website, every worker and Redis as it happens. Filter, show errors only, pause, download.

Resource graphs

CPU, memory and network for each project over an hour, a day or a week, sampled every minute.

Instant rollback

Releases switch only after a health check passes. Earlier ones stay ready, so going back is one click, not a rebuild.

New · AI assistant

A build failed?
Claude tells you why.

One click on a failed deployment. Claude reads the end of the build log and your settings, then says what broke, why, and exactly what to change, with how sure it is.

  • Every variable’s value is replaced by its name before anything is sent
  • Your own Anthropic key, entered once by an admin, stored encrypted
  • Saved with the deployment, so your team reads it for free
Failed#47 · Upgrade to Node 22

Next.js could not load its compiler on Alpine. high confidence

The node:22-alpine image needs the musl build of SWC, which your lockfile doesn’t include because it was made on macOS.

  1. Run npm i -D @next/swc-linux-x64-musl and commit the lockfile.
  2. Or set the base image to node:22-slim in your Dockerfile.
  3. Redeploy.

Explained by Claude · secrets hidden

The real thing

One dashboard for the whole server.

Screens from DeployPro itself: the Overview with the server’s memory and CPU and every project’s share, and a project’s live logs.

app.deploypro.us/overview
The DeployPro Overview: projects, memory in use, CPU load, memory claimed, the projects using the most memory, and each project's limits.
app.deploypro.us/projects/blog/runtime/logs
A project's live runtime logs in DeployPro, with errors highlighted and a filter.

Platform

Everything a production app needs.
On one server.

Domains with automatic HTTPS

Add a domain; DeployPro creates the DNS records through Cloudflare and the certificate through Let’s Encrypt.

Workers and cron jobs

Queue consumers, schedules and background services beside your site, with replicas, pause and resume.

Built-in Redis

A private Redis per project for BullMQ, Celery or Sidekiq, kept running and backed up.

A team, safely

Viewer, developer, admin and owner roles, two-factor sign-in, API tokens and an audit log of every change.

Alerts where you work

Deploy results and outages to Slack, Discord or Microsoft Teams, per project.

Backups that leave the server

Daily backups of the database, volumes and Redis, with an off-site copy to a Hetzner Storage Box.

Secrets handled properly

Variables are encrypted at rest, passed to builds as secret mounts that never reach an image layer, and shown only when someone with the right role asks. Paste a whole .env to move from another host in one go.

Ownership

The cloud experience,
on hardware you control.

Rent one server from Hetzner or anyone else. Run your sites, APIs, workers and queues on it with the workflow you’d expect from a hosted platform, and keep the data, the logs and the bill.

Servers to start
1
Per-seat fees
0
Build-minute fees
0
Where your data lives
Yours

Self-host

Up in one command.

On a fresh Ubuntu or Debian server with a domain on Cloudflare. The installer sets up Docker, the secrets, the firewall, HTTPS and the dashboard, then checks everything works. Run it again to upgrade.

  1. Point DNS at the server: a wildcard for deployments.
  2. Run the installer with your domain and DNS token.
  3. Sign in, connect GitHub, import a repository, push.
root@server
# on the server
git clone https://github.com/ICOFCUCAM/deploygenus.git /opt/deploypro
cd /opt/deploypro
DEPLOYPRO_DEPLOY_DOMAIN=deploys.example.com \
DEPLOYPRO_ACME_EMAIL=you@example.com \
CF_DNS_API_TOKEN=… bash scripts/install.sh

✓ Docker ready      ✓ TLS issued
✓ Database migrated ✓ doctor: all clear

Push. It’s live. On your server.

Already running DeployPro? Your dashboard is at app.deploypro.us.