- add force_domain_override functionality and docs
- delete service on creation if there is URL conflicts as otherwise we will have stale services (we need to create the service because we need to parse it and more)
- remove `docker_compose_raw` from post and patch endpoints, as the compose file is sourced from git and should not be manually settable via the api
- improve the documentation for `docker_compose_domains` (URLs)
- enhanced array validation for `docker_compose_domains` by validating each array field and verifying which fields are allowed
- set a custom array validation error message, as the default message is not really clear
- show an error if the user attempts to set `domains` when the build pack is `dockercompose`
- validate that the `domains` in `docker_compose_domains` are proper URLs and include a valid scheme (`http` or `https`)
- if service type and docker_compose_raw is filled show an error
- if service type is not valid show an error with all valid service types
- remove enum from service type docs as it always gets outdated
Replace hardcoded 'debian' with ${ID} from /etc/os-release to use
the correct Docker repository for Ubuntu, Debian, and Raspbian servers.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replace insecure !== operator with hash_equals() for constant-time
string comparison when validating GitLab webhook tokens.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- raw sql should not be used whenever possible
- using sql to order on the relationship environment_variables() causes custom sorting to break or be additionally complicated
Remove the redundant $original_server property and use $mainServer throughout
ApplicationDeploymentJob. Both properties held the same value (the deployment
target server), causing unnecessary duplication.
Also fixes two bugs in generate_compose_file() where $this->server was used
instead of $this->mainServer for isSwarm() and isLogDrainEnabled() checks.
When using a build server, $this->server could point to the build server,
causing incorrect configuration for the deployment target.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The instantSave method for the 'Make it publicly available' checkbox was
calling submitDatabase(), which saved all form fields. This caused
unintended saves when only toggling the public visibility. The
syncDatabaseData() call already handles saving the public toggle state.
The proxy container was incorrectly named using the service UUID instead
of the database UUID, causing proxy logs to query the wrong container.
Each ServiceDatabase should have its own uniquely named proxy container.
- Introduced a new sidebar component for service database navigation.
- Updated routes for database import and backup functionalities.
- Refactored the database import view to improve clarity and maintainability.
- Consolidated service application and database views into a more cohesive structure.
- Removed deprecated service application view and integrated its functionalities into the service index.
- Enhanced user experience with modal confirmations for critical actions.
- Improved code readability and organization across various components.
Replace in-memory filtering with database-level query in
Server::destinationsByServer(). Previously loaded all team servers
into memory before filtering by ID. Now uses findOrFail() to query
directly at the database level.
Add email observability platform Sessy to the service catalog with auto-generated HTTP Basic Auth credentials and SQLite storage.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Wraps inline chart initialization scripts in IIFEs to create local scope for variables. This prevents "Identifier has already been declared" errors when Livewire's SPA navigation re-executes scripts, allowing smooth navigation between metrics pages without page refresh.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
The trim operation was happening after validation, which meant
whitespace was counted toward the max:255 validation rule. Now
input is normalized before validation, matching the pattern used
in Application and Service components.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Andras Bacsai <andrasbacsai@users.noreply.github.com>
Add visual feedback when downloading all logs in both container and deployment log views. Users now see an animated spinner and "Downloading..." text, preventing multiple concurrent downloads and improving UX during long operations.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Remove unused $server property and Server import from Advanced.php
- Add proper import for UpdateStripeCustomerEmailJob in User.php
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add trim() to FQDN in instantSave() method to prevent whitespace from being saved
- Add trim() to FQDN in submit() method before validation and DNS checks
- Prevents invalid HostSNI rules caused by leading/trailing whitespace
- Fixes issue where accidental whitespace from copy-paste causes deployment failures
Fixes#7797
Co-authored-by: Andras Bacsai <andrasbacsai@users.noreply.github.com>
- Adjusted indentation and formatting for better code clarity.
- Simplified conditional statements and removed unnecessary null checks.
- Enhanced user interface elements for Docker Compose and Dockerfile configurations.
- Improved handling of deployment commands and network settings.
- Updated helper texts for better user guidance.
Users can now choose between downloading only the currently displayed
logs or fetching and downloading all available logs from the container.
Changes:
- Add downloadAllLogs() method that fetches all logs without limit
- Replace download button with dropdown menu
- Options: "Download displayed logs" and "Download all logs"
Addresses #7803
The log viewer was artificially limiting display to 2000 lines
regardless of user's requested amount. Users could request 10k, 40k,
or 50k lines but only 2000 were ever shown.
Changes:
- Remove the hardcoded $maxDisplayLines = 2000 limit in the view
- Add MAX_LOG_LINES constant (50,000) in GetLogs component
- Enforce maximum limit in backend to prevent extremely large requests
- Update input field with max attribute and tooltip
Fixes#7803
- Include sessionTeamId in currentTeam() cache key to prevent stale
team data when users switch teams
- Update refreshSession() to use new cache key format
- Remove redundant routeIs('settings.index') check since settings.*
already matches it
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Fix currentTeam() to return null instead of crashing when no session
- Fix role() to use $this->currentTeam() instead of global helper
- Add roleInTeam() method for explicit team context
- Remove unused otherTeams() method
- Fix InviteLink authorization bypass when role() returns null
- Fix confirmEmailChange() null safety for currentTeam()
- Fix ActivityMonitor to handle null currentTeam with fallback chain
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Fix isInstanceAdmin(), currentTeam(), otherTeams(), and role() methods
to operate on the actual User instance instead of always using the
authenticated user. This ensures correct behavior when these methods
are called on non-authenticated user instances (e.g., in ActivityMonitor).
Also fix settings route check to use routeIs() instead of path matching.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Make Server property nullable in Settings components (Index, Advanced, Updates)
- Add conditional server loading: only load when not on cloud
- Add null checks before using server for DNS validation and proxy configuration
- Fix isInstanceAdmin() to check root team's pivot role directly instead of current team
- Make root team (id=0) bypass subscription check on cloud
- Remove isInstanceAdmin() from main middleware bypass: only settings/admin routes are exempted
- Update isSubscribed() to only check isSubscriptionActive() for navbar consistency
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Add smart goBack() method that skips auto-selected steps and returns to the
last step where user had a real choice. This prevents navigation loops when
previous steps only have a single option and auto-select.
Fixes#7739🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Wrap return values in collect() to maintain Collection compatibility
- Add comment explaining threshold <= 2 prevents division by zero
- Refactor tests to use actual Server model method via reflection
- Use seeded mt_rand() for reproducible test results
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implement the Largest-Triangle-Three-Buckets (LTTB) algorithm to downsample
metrics data for large time intervals (30 days generates 260K-500K+ points).
Reduces rendered points to ~1000 while preserving visual accuracy of peaks
and valleys. Fixes unresponsive page when selecting 30-day metrics interval.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
When a database or application was in a restart loop, the restart count
persisted even after the user manually stopped the resource. This caused
the UI to continue showing "(Xx restarts)" after user intervention.
Now resets restart_count, last_restart_at, and last_restart_type when:
- User stops a database (StopDatabase action)
- User stops an application (StopApplication action)
The existing reset in GetContainersStatus is still needed for containers
that exit on their own (crash without recovery, Docker giving up).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Fixes#7732 - The proxy status change listener was dispatching ProxyStatusChangedUI
before the Traefik version check job had a chance to run. This caused users to see
stale version information when they refreshed the page immediately after restarting
the proxy.
The fix defers the UI refresh when a Traefik version check is being dispatched. The
version check job already dispatches its own ProxyStatusChangedUI event when
complete, ensuring the UI refreshes with updated version data.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Add public instantSave() method to handle instant saves from checkbox clicks
- Remove redundant updatedIsMetricsEnabled() and updatedIsSentinelDebugEnabled() hooks
- These hooks were causing duplicate notifications when checkboxes were toggled
The instantSave attribute on checkboxes triggers wire:click='instantSave', which was failing
because the method didn't exist. Now it saves settings and restarts Sentinel in one action,
preventing the duplicate updates from both wire:click and wire:model events.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Add cloud:sync-stripe-subscriptions command to manually check all
subscriptions against Stripe. By default it only reports discrepancies
without making changes. Use --fix flag to actually apply corrections.
This addresses race conditions where subscriptions can be cancelled in
Stripe but remain marked as active in Coolify's database.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Builder containers are started with the --rm flag, which automatically removes them when stopped. The explicit docker rm -f is redundant and adds unnecessary steps to deployment logs.
This change adds a skipRemove parameter to graceful_shutdown_container() and sets it to true for builder container shutdowns (uuid-based) while keeping the default behavior for application containers.
Fixes#7566🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Helper containers are started with --rm flag which automatically removes the container when it stops. Removed redundant docker rm commands from graceful_shutdown_container in ApplicationDeploymentJob and replaced docker rm with docker stop in DatabaseBackupJob.
🤖 Generated with Claude Code
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Remove per-line x-effect directives that re-evaluated for every log line during polling
- Replace with efficient applySearch() function that updates logs once after Livewire morph
- Remove unnecessary caching mechanisms (renderTrigger, decodeCache, matchCountCache)
- Remove double HTML encoding of log lines (e() + Blade escaping)
- Add decodeHtml() helper to properly decode HTML entities from data attributes
- Use morph.updated hook instead of commit hook for efficient DOM updates
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Create new Server/Swarm.php Livewire component and view for Swarm configuration
- Create new Server/Sentinel.php Livewire component and view for Sentinel settings
- Add server.swarm and server.sentinel routes
- Move Swarm and Sentinel sections from General page to sidebar menu items
- Improve organization by separating concerns into dedicated pages
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Use atomic update pattern in backend: collect logs into temp variable
before replacing outputs (prevents empty state flash)
- Remove per-line x-effect directives that caused 4000+ reactive
evaluations on every update
- Replace inline Alpine.js class bindings with CSS utility classes
- Use single $watch and morph.updated hook instead of renderTrigger
- Remove HTML entity decode cache (no longer needed)
- Fix search highlight padding that caused text shifting
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add missing edge case test for root path (/) and expand test coverage
for preview FQDN generation. Tests verify that ports and paths are
correctly preserved in preview URLs while excluding root paths.
Fixes#2184
Add copy-to-clipboard functionality for both deployment logs and runtime container logs with success notification. Fixed event dispatch to use Livewire.dispatch() for proper toast notification handling. Reorganized copy and download buttons to appear consecutively in runtime logs toolbar.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
The Resources tab threw a "Queueing collections with multiple model types is not supported" error because the Livewire component was storing a mixed-type Eloquent collection (Applications, Databases, Services) as a public property, causing Livewire's serialization to fail.
Fixed by: storing only the unmanaged containers array in the component, and calling definedResources() directly in the Blade view for the managed tab.
Fixes#7666🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Consolidate duplicate restart tracking logic in GetContainersStatus
- Add last_restart_type string cast to all 8 standalone database models
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Fix status flickering: Track databases in active/transient states (restarting, starting, created, paused) not just running
- Add isActiveOrTransient() helper to distinguish between active states and terminal states (exited, dead)
- Add safeguard: Protect updateNotFoundDatabaseStatus() from marking as exited when containers collection is empty
- Add restart_count tracking: New migration adds restart_count, last_restart_at, last_restart_type to all standalone database tables
- Update 8 database models with $casts for new restart tracking fields
- Update GetContainersStatus to extract RestartCount from Docker and update database models
- Reset restart tracking when database exits completely
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Allow users to select PostgreSQL version instead of automatically creating postgres:16-alpine when using global search. The fix includes:
- Remove hardcoded database_image parameter from GlobalSearch
- Update Create.php to fall through to Select component when database_image is not provided
- Add type and destination to Select component query string with proper URL mapping
- Jump directly to PostgreSQL version selection step when navigating from global search
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Implement instance-wide SPA navigation toggle that enables smooth page transitions with prefetching on hover. Excludes terminal links which require full page lifecycle for WebSocket connections. Adds defensive checks to global-search component for SPA navigation compatibility.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Set modal width to consistent 48rem for both upgrade states
- Remove max-width constraint from progress stepper
- Add dev mode with Simulate button for local testing
- Simulate cycles through all upgrade steps with 2-second delays
- Removed unnecessary SVG icons from the environment edit view for cleaner UI.
- Deleted the environment select component as it was no longer needed.
- Enhanced the project resource index view with a dropdown for environments and resources, improving navigation.
- Implemented dynamic dropdowns for environments and their associated resources, allowing for better user interaction.
- Added transitions and hover effects for a more responsive design.
- Updated the layout to ensure a consistent user experience across different project resources.
Add a copy button to individual container logs that strips sensitive
data before copying to clipboard. Includes sanitization for emails,
database URLs with passwords, JWT tokens, API keys, private key blocks,
and git access tokens.
- Eager load service applications and databases to eliminate N+1 queries
- Replace individual model updates with batch database updates for applications, previews, and services
- Move connectProxyToNetworks to async ConnectProxyToNetworksJob to avoid blocking status updates
- Optimize Server.databases() and applications() methods with efficient database queries
- Use flatMap for cleaner collection transformations
🤖 Generated with Claude Code
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Add model saving event to trim key/secret fields (encrypted casts)
- Add attribute mutators to trim endpoint, bucket, region fields
- Create migration to fix existing S3 storage records with whitespace
- Use chunking in migration to handle large datasets efficiently
- Verify re-encryption validity before committing changes
Fixes#7469#7594🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Prevent text selection from being cleared when logs are re-rendered during polling
- Preserve fullscreen state when toggling debug logs or other Livewire updates
- Fix log filtering to properly apply when debug mode is toggled
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Fix null-safe operator on currentTeam() call in Upgrade.php
- Add --rm flag to docker run in upgrade.sh for cleanup consistency
- Store beforeunload handler as named reference and clean up on success
- Add clarifying comments for upgrade method calls
- Add error state handling with close option in upgrade modal
- Add step mapping documentation comment in upgrade-progress component
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When build server is enabled, $this->server points to the build server.
The log drain configuration check was using $this->server which would
incorrectly check the build server's settings instead of the deployment
server where the container actually runs.
This fix ensures log drain configuration is correctly applied based on
the deployment server's settings by using $this->original_server.
- Remove /api/upgrade-status endpoint and route
- Add getUpgradeStatus() method to Upgrade Livewire component
- Update frontend to call Livewire method via $wire.getUpgradeStatus()
- Simpler approach: no separate API, auth handled by Livewire
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Temporary debug fields added to identify why status returns 'none'
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The status file is on the host filesystem, not inside the container.
Use instant_remote_process() to read the file via SSH to Server::find(0).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add auth:sanctum middleware to /api/upgrade-status route
- Check user belongs to root team (id 0) before returning status
- Return 403 if user is not authorized
- Update frontend to send credentials with fetch request
- Update OpenAPI docs with 401/403 responses
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Delete status file 10 seconds after upgrade completes
- Reduce stale timeout from 30 to 10 minutes
- Remove timestamp from API response (internal detail)
- Treat timestamp parse failures as stale for security
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- upgrade.sh now writes status to /data/coolify/source/.upgrade-status
- New /api/upgrade-status endpoint reads status file for real progress
- Frontend polls status API instead of simulating progress
- Falls back to health check when service goes down during restart
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add step-by-step progress indicator (Preparing → Helper → Image → Restart)
- Display elapsed time during upgrade (MM:SS format)
- Show version transition in header (v4.0.0-beta.454 → v4.0.0-beta.456)
- Add expandable changelog preview before upgrading
- Reduce reload delay from 5s to 3s with countdown timer
- Add "Reload Now" button to skip countdown
- Improve status messages with step-specific descriptions
- Add success state with clear indication when upgrade completes
- Create new upgrade-progress component for visual step tracking
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
- Show clear progress with numbered steps (1/6 through 6/6)
- Display header and footer banners
- Show individual image pull progress
- Show which containers are being stopped
- Display final success message with version and log location
- Keep detailed logging to file for debugging
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
OAuth users don't have passwords set, so they should not be prompted for password confirmation when performing destructive actions. This fix:
- Detects OAuth users via the hasPassword() method
- Skips password confirmation in modal for OAuth users
- Keeps text name confirmation as the final step
- Centralizes logic in helper functions for maintainability
- Changes button text to "Confirm" when password step is skipped
Fixes#4457🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Filter webhook-triggered deployments by source_id to ensure only applications
associated with the GitHub App that sent the webhook are deployed, preventing
duplicate deployments when the same repository is configured in multiple teams.
Test emails should work with any recipient email address for verification purposes, not just team members. Added an isTestNotification flag to both Test notification classes and modified EmailChannel to skip team membership validation for test notifications.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
Allow manually-added servers to be linked to Hetzner Cloud instances by
matching IP address. Once linked, servers gain power controls and status
monitoring.
Changes:
- Add getServers() and findServerByIp() methods to HetznerService
- Add Hetzner linking UI section to Server General page
- Add unit tests for new HetznerService methods
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add isServiceResource() and shouldBeReadOnlyInUI() to LocalFileVolume
- Update path matching to handle leading slashes in volume comparisons
- Update FileStorage and Show components to use shouldBeReadOnlyInUI()
- Show consolidated warning message for service/compose resources in all.blade.php
- Remove redundant per-volume warnings for service resources
- Clean up configuration.blade.php formatting
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Use relationLoaded() check before accessing the application relationship
to avoid triggering individual queries for each volume when rendering
storage lists. Update Storage.php to eager load the relationship.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Fixed isReadOnlyVolume() to detect both short-form (:ro) and long-form (read_only: true) Docker Compose volume syntax
- Fixed path matching to use mount_path only (fs_path is transformed during parsing from ./file to absolute path)
- Added "Load from server" button for read-only volumes to allow users to refresh content
- Changed loadStorageOnServer() authorization from 'update' to 'view' since loading is a read operation
- Added helper text to Content field warning users that content may be outdated
- Applied fixes to both LocalFileVolume and LocalPersistentVolume models
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Current version of infrastructure images (coolify-helper, coolify-realtime) are now protected from deletion during docker cleanup, regardless of which registry they're pulled from (ghcr.io, docker.io, or Docker Hub implicit). Old versions continue to be cleaned up as intended.
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
- Add 429 response with Retry-After header for Hetzner server creation
- Create RateLimitException for proper rate limit error handling
- Rename cloud_provider_token_id to cloud_provider_token_uuid with deprecation
- Fix prices array schema in server-types endpoint with proper items definition
- Add explicit default: true to autogenerate_domain properties
- Add timeout and retry options to Docker install curl commands
- Fix race condition in deployment status update using atomic query
Adds Retry-After: 60 header to all deployment queue full responses,
helping webhook clients know when to retry their requests.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Remove unused server filtering logic in Kernel.php that was querying servers
but never using the results. Simplify Sentinel update checks in ServerManagerJob
by reusing the $isSentinelEnabled variable and removing unnecessary timezone
parameter for hourly cron execution.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Prevent deployment status from regressing to FAILED after it's marked as FINISHED by:
1. Calling completeDeployment() first in post_deployment() before any operations that could fail
2. Wrapping all post-deployment side effects in try-catch blocks
3. Adding FINISHED to terminal states that cannot be changed
4. Protecting ExecuteRemoteCommand from overwriting FINISHED status
This fixes the issue where a deployment with a healthy container and successful rolling update was still marked as Failed in the UI.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
