On self-hosted, a validation that connects now clears force_disabled. The unreachable cleanup skips the localhost server, and the server page explains why a server is disabled. Cloud and transferred servers stay disabled.
Closes#12475
- Add a hint to failed command messages when the exit code is 255,
which explains that the SSH connection to the server closed before
the command finished.
- Return no services from applicationParser and serviceParser when the
compose `services` key is empty or is not a map, so parsing does not
fail.
- Move Server::getIp() out of the sshEndpointInUse() docblock, where it
was misplaced and broke the class.
- Add tests for the SSH hint, for services keys that are not maps, and
for saved compose variable values that must survive a redeploy
(#6995).
caddy-docker-proxy reloads Caddy whenever any proxied container changes,
and Caddy closes every WebSocket on reload by default. Set
stream_close_delay on app, service, and dashboard reverse_proxy routes.
Closes#7942
Each new SSH connection opens a systemd-logind session. Some systemd
versions leak closed sessions until the host refuses logins at 8192.
The connection check now opens one new login (the probe), not three.
Closes#9148
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Deleting a server with "Delete all resources" (UI) or ?force=true (API)
deleted every application that used the server, also applications that
use it only as an additional destination, for example after Promote.
Server delete now deletes only resources whose main destination is on
the server. Applications that use it only as an additional destination
lose their additional_destinations row for the server, and their
containers on it are stopped (best effort). The check for resources and
the transfer archive path use the same Server::primaryResources().
The delete confirmation lists the resources to delete and the
applications that keep running on another server.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The browser uses only the /app/{key} WebSocket. Laravel sends events to
Reverb over 127.0.0.1, so the generated Traefik and Caddy configs no longer
route /apps to port 6001, and the WebSocket rule only matches /app/.
Closes#12390
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Add an Axiom edge domain setting (default us-east-1) and use the /v1/ingest endpoint, so EU datasets work
- Remove the Highlight drain (service ended) and stop its running containers
- Update Fluent Bit from 2.0 to 5.1 and use the warn log level
- Restart running drains when the generated files change, with the new image pulled before the old container stops
- Show log drain validation errors under their fields
Closes#9898
A user database named coolify-db was backed up as the instance database,
which targeted the wrong container. It was also hidden from the server
database list and from status updates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Refresh the navbar proxy status on the 30 s poll
- Skip Sentinel and server check auto-start while a proxy start or restart runs
- Use the StartProxy commands in RestartProxyJob, so Caddy restarts keep the
maintenance blocks and remove the Traefik logrotate sidecar
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
When "docker version --format json" did not return clean JSON (for
example, text from a shell startup file), validation stored 0.0.0 and
said that Docker was older than version 24. Validation now stores no
version and tells the user to run the command on the server. When the
version is too old, the message shows the detected version.
Closes#11089
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Import:
- Check every uuid in a bundle against a strict id format before any
write, and quote the SSH mux socket and key paths in local shell
commands. A bundle uuid could run commands in the Coolify container.
- Validate bundle fqdns, git source URLs, paths, networks, cron
expressions, and database ports; cap the number of items.
- Do not confirm that another team has a uuid.
- Refuse a project icon path with a uuid that is not a plain id.
Claim, finish, and migrate:
- Pass checklist step keys to wire:click with Js::from (stored XSS).
- Link only to https target server URLs.
- Validate target_instance_url in /transfer/complete; never prefill an
unverified URL for a transfer back; scope the transfer-back token to
write+deploy for 1 hour.
- Release on an owner marker only when it is newer than the own claim.
- Do not touch apps that only use the server as an additional
destination when an archive is released or deleted.
- Give Sentinel a new token on claim.
- Add a checklist step to remove the previous instance's SSH key.
- Block a transfer while backups or scheduled tasks run.
- Authorize refreshMigration and PrivateKey checkConnection, clear
tokens from Livewire state, hide migration log/error from members.
Archives:
- Refuse ip, user, port, private key, and Cloudflare tunnel changes.
- Throttle Enable management and require a free server slot for it and
for a transfer back.
- Skip cloud provider calls.
- Re-read force_disabled before each SSH command, so running jobs stop.
- Probe the import check target at the resolved address only.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Rate limit the import check probe and refuse private, loopback, and
link-local targets on Cloud.
- Rate limit the SSH validation that an import starts, like other
server validations.
- Count only managed servers for the server limit, the limit check job,
and the subscription page. Archives of transferred servers are left out.
- Do not show the name of another team's archived server in the import check.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Skip TrimStrings and ConvertEmptyStringsToNull on the import route; an empty shared variable failed the whole import
- Keep compose parser versions, static ports, legacy variable keys, and only the bundle's variables and destinations
- Give the transfer back token the deploy permission that the import needs
- Move the Cloudflare tunnel DNS record with the server and park it on the source
- Refuse cloud server deletes, power on, tunnel and DNS changes, and scp for archives; skip their backup alerts and provider polling
- Add a column-level round-trip test, archive guard tests, and a live full-coverage transfer test
- Advance the GitLab App sequence in the dev seeder
- Build a valid compose device request (device_ids list, options mapping, blank driver)
- Map --gpus N to count and drop --gpus when GPU settings are enabled
- Validate only changed GPU values so stored legacy values keep working
- Detect NVIDIA GPUs and the container toolkit with the server details
- Show detected GPUs with a refresh button in the application GPU section
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Applications and service applications can now use the server resolver,
no resolver ("none"), or a custom resolver name. This allows servers
that mix Let's Encrypt domains with domains that use their own
certificates, such as Cloudflare Origin CA. The setting is available in
the UI and the API.
Closes#12138
Coolify always added tls.certresolver=letsencrypt to generated HTTPS
routers, so Traefik logged errors when the proxy used its own
certificates (for example Cloudflare Origin CA) without that resolver.
Add a server setting (default letsencrypt) in the proxy page and API.
An empty value omits the resolver from resource labels, the Coolify
dashboard routes, and the catch-all router.
Closes#12138
Store server IPs in one form (no brackets, short lower-case IPv6) and
write each address in the form its context needs:
- URLs and host:port (Sentinel, database URLs, SFTP backup download,
DNS servers, SMTP, webhook base URLs, links, health checks) get
brackets through formatHostForUrl().
- Traefik Host() rules use the address without brackets, so IPv6
domains no longer return 404; saved labels are fixed at deploy.
Caddy keeps the brackets.
- SSH keeps the plain address.
- Hetzner gives an IPv6 /64 network; IPv6-only servers now get the
server address (prefix::1) instead of the unusable prefix::64.
- sslip labels, registry image names with an IPv6 host, the API domain
list, allowlist matching and duplicate checks handle IPv6.
A migration normalizes stored IPv6 server addresses, repairs Hetzner
servers saved as ::64 that are not reachable, and adds brackets to
saved Sentinel URLs.
Queued deployments of a deleted server could never start and stayed in
the queue. Deleting a server now marks them failed with the log entry
"The server was deleted."
Allow validated host file and directory mounts outside resource directories
while protecting external paths from deletion.
Clean up cancelled deployment containers and clear tracked process IDs.
Preserve legacy KeyDB and Dragonfly password parsing, accept server backup
paths without extension restrictions, and allow Compose variable proxy ports.
- Restart a running Sentinel that has not pushed for 10 minutes, at most
once per hour. The restart applies the current URL and token.
- Restart an unhealthy Sentinel. Do not restart while its health check
starts or while its API does not answer yet, which caused restart loops.
- Remember why pushes fail: the reason Coolify rejects a push, and the last
push error from Sentinel's /api/push-status. Show it on the Sentinel page
while Sentinel is out of sync.
- When the instance URL or public IP changes, move remote servers that use
the generated Sentinel URL to the new URL. Custom URLs stay.
- ServerCheckJob uses the stored reachability state instead of a second SSH
probe. One failed probe no longer marks the server unreachable without
UNREACHABLE_THRESHOLD or marks all its resources exited.
- Do not dispatch ServerCheckJob for build servers and Swarm workers.
- Check cloud provider state every 5 minutes instead of every minute.
- ServerPatchCheckJob uses isFunctional() and drops the unused retries.
- Delete the unused ServerCleanupMux job.
- Disable SSH multiplexing per command. The global config change stayed in
the queue worker, so later jobs (also deployments) ran without mux.
- Start Sentinel with --restart unless-stopped so Docker restarts it
after a crash or reboot.
- Make CheckAndStartSentinelJob unique per server so copies cannot
collect in a slow queue.
- Read X-Sentinel-Version on push and update an outdated Sentinel
without SSH; skip the hourly SSH version check for servers that
report their version.
Deployments failed with "Server is not functional" when the server was
still flagged unreachable after a single failed scheduled check. Add
Server::isFunctionalAfterRecheck(), which runs a live SSH and Docker
connection check when the server is marked unreachable. It skips the
check for disabled servers and servers with placeholder IPs, then
re-evaluates isFunctional(). ApplicationDeploymentJob now uses it.
ServerManagerJob no longer skips the SSH connection check for servers
with a live Sentinel when the server is flagged unreachable or unusable.
The Sentinel heartbeat does not restore these flags, so without the
check they would never recover.
Add feature tests for the recheck behavior and unit tests covering the
Sentinel skip condition.
A single failed SSH attempt or any unrelated exception in
ServerConnectionCheckJob marked the server unreachable and unusable, so
deployments failed with "Server is not functional" while the server was
online.
- Retry a failed SSH attempt once after 3 seconds within the same check.
- Mark the server unreachable only after UNREACHABLE_THRESHOLD (2)
consecutive failed checks; notifications use the same threshold.
- Keep the stored state when the job fails for a reason other than SSH or
Docker (database, cache, bugs).
- Raise the job timeout to 60 seconds and the overlap lock above it. The
old 15 second timeout killed checks before they wrote a result.
- Run the Docker and Compose version commands once with a 15 second
timeout. A failed Compose check no longer marks the server unusable.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Regenerate SSL Certificates now writes server.crt/server.key for every
engine except MongoDB, which keeps the combined server.pem. Before, the
button always wrote server.pem, so Postgres, MySQL, MariaDB, Redis, KeyDB
and Dragonfly failed to start after a regenerate.
- The certificate renewal job uses the same rule, so MongoDB no longer
switches to server.crt/server.key on renewal and fails on restart.
- Starting an SSL database writes the CA that signs its certificate to
/data/coolify/ssl/coolify-ca.crt, so a stale CA file (for example one
written by another server entry for the same host) no longer causes
"certificate signature failure".
- Saving a custom CA certificate that does not match the stored CA
private key is rejected.
- The CA file write commands live in SslHelper::caCertificateFileCommands().
Fixes#10388Fixes#6771Fixes#11926
- database import: add keep_owners (restore PostgreSQL owners and
privileges) and sqlite_database (pick the target SQLite file) to the
API, OpenAPI spec, and import form
- oauth: require password confirmation and support team selection on
OAuth login
- remote process: send team broadcast events to the team of the server
that ran the process instead of the causer's user id
- servers API: handle the legacy build server setting
- compose: support parent bind paths and volume driver options
- dns: improve record hints and provider management
- notifications: fix throttling and missing backup notifications
- ui: rework the confirmation modal
- add feature and unit tests for the new behavior
Andras Bacsai
