Webhook secret was stored and shown as plaintext. Use a password field,
encrypt at rest (with legacy plaintext read support), and look up tokens
via findByWebhookToken so encrypted values still authenticate webhooks.
Match application status layout by placing the Connected badge next to
the page heading, and drop the redundant Disconnect action (re-auth is
done via Connect after tokens expire or credentials are updated).
Match the GitHub App endpoint picker so self-hosted / tunnel setups can
select FQDN, IP, app URL, or a custom base. Redirect URI is derived as
{base}/webhooks/source/gitlab/redirect and persisted for token exchange.
The secret was always stored encrypted, but the setup form wiped the
input on every load. Load it back for admins (GitHub App parity) so a
reload no longer looks like a failed save.
Use the red incomplete-setup alert like GitHub, keep name + OAuth
credentials front-and-center, and tuck GitLab URL / API / SSH / system
wide options into an Advanced accordion for self-hosted users.
Match the GitHub create modal layout: intro copy, name/group row,
system-wide warning, self-hosted accordion (URL/API/SSH), and a
bottom Continue button instead of a duplicate header Save.
GitlabAppPolicy previously allowed any authenticated user to update,
delete, and create GitLab sources. Align it with GithubAppPolicy, require
Application create authorization on the private-repo wizard, and reject
OAuth callbacks from non-admins so members cannot escalate privileges.
Adds self-hosted GitLab OAuth sources so Coolify can connect to a self-managed GitLab instance, list private repositories, clone over an OAuth token, and deploy (the GitLab counterpart to GitHub Apps).
Hardening: authenticated, one-time team-bound OAuth callback state; token redaction in deploy logs; custom host port/path kept in clone and ls-remote URLs; submodule OAuth auth; system-wide source selection. Covered by unit and feature tests.
cosigned by OpenAI Codex at M1 Max
- Gate volume backup retention and S3 controls by update permission
- Preserve backup records when S3 deletion fails
- Share SFTP download streaming with consistent missing-file handling
- Handle schedule creation errors and link service database backups
Introduce V5Feature and config so V5 routes, jobs, commands, morph maps,
and model queries run only when enabled. Move V5 migrations to
migrations-v5 (loaded only when enabled), remove Flux from production
Docker/install paths, and add isolation tests.
Restrict V5 application and resource-connection creation to team admins.
Resolve project, environment, and application from query params and keep
session selection in sync. Surface V5 apps on the v4 resource index and
count them for project/environment emptiness. Create the flux data dir
on install and upgrade.
Expose DELETE for application, database, and service storage backup
schedules (with OpenAPI docs), reject storage deletes while schedules
exist, skip retention cleanup when no limits are set, and remove S3
archives using the execution’s S3 storage.
Store s3_storage_id on scheduled volume backup executions so retention
and recovery use the S3 that received the upload. Extract
DeleteScheduledVolumeBackup for UI and resource deletion, and isolate
database backup retention failures so cleanup errors do not fail the job.
Wrap Vultr server creation in DB transactions and delete the remote
instance when local persistence fails (API and Livewire). Scope
plus/dot email normalization to gmail.com/googlemail.com only.
Use throw:false on DigitalOcean/Vultr HTTP retries, and normalize
service log line counts via normalizeLogLines.
Enable scheduled ClickHouse backups across the job, API, and UI, and
guard unsupported database types via isBackupSolutionAvailable().
Convert Stripe subscription sync from a job to an action with clearer
discrepancy resolution, and add cloud:export-users plus
cloud:cleanup-unverified-users with tests.
Add polymorphic volume backup scheduling for persistent volumes and directories, expose schedule management via API, and reorganize backup configuration and execution views.
Persist DigitalOcean, Hetzner, and Vultr servers before public IP
assignment, then backfill placeholder addresses from provider state.
Treat partial Sentinel snapshots as non-authoritative and document the
destinations API with OpenAPI schemas.
Add server private key generation from the dropdown, prevent deleting
private keys that are still in use, and close cloud provider link modals
after successful linking.
Add dedicated show/edit pages for cloud provider tokens and cloud-init
scripts, including descriptions and UUID routes.
Generate private keys directly from the index and surface cloud provider
API loading errors in server creation flows.
Add normalized email identity rate limiting for registration and forgot-password requests, and refresh Sentinel status from restart broadcasts.
Rework server sidebars and navbar for mobile menus and active status visibility.
Require IPv6 when public IPv4 is disabled, constrain cloud token
lookup by team and provider during deletion, and encode Vultr instance IDs
in API paths.
Authorize stored Hetzner cloud provider tokens before fetching
firewalls or networks, hide provider error details, and keep server
creation intact if backup activation fails.
Collapse advanced Hetzner server options by default while preserving
visibility when advanced values are selected.
Normalize tag names before attaching them, reject names that are too short
after sanitization, and return 404 when removing tags not attached to the
resource.
Adds a per-team unique tag-name index and migrates duplicate tags onto the
kept record before creating the constraint.
Andras Bacsai
Mike ChongandAndras Bacsai
