Fix Docker cleanup irregular scheduling in cloud environment

Description

Docker cleanup jobs are running at irregular intervals instead of hourly as configured (0 * * * *) in the cloud environment with 2 Horizon workers and thousands of servers. The issue stems from the ServerManagerJob processing servers sequentially with a frozen execution time, causing timing mismatches when evaluating cron expressions for large server counts.

Acceptance Criteria

<!-- AC:BEGIN -->

• #1 Docker cleanup runs consistently at the configured hourly intervals
• #2 All eligible servers receive cleanup jobs when due
• #3 Solution handles thousands of servers efficiently
• #4 Maintains backwards compatibility with existing settings
• #5 Cloud subscription checks are properly enforced

<!-- AC:END -->

Implementation Plan

1. Add processDockerCleanups() method to ScheduledJobManager

   • Implement method to fetch all eligible servers
   • Apply frozen execution time for consistent cron evaluation
   • Check server functionality and cloud subscription status
   • Dispatch DockerCleanupJob for servers where cleanup is due

2. Implement helper methods in ScheduledJobManager

   • getServersForCleanup(): Fetch servers with proper cloud/self-hosted filtering
   • shouldProcessDockerCleanup(): Validate server eligibility
   • Reuse existing shouldRunNow() method with frozen execution time

3. Remove Docker cleanup logic from ServerManagerJob

   • Delete lines 136-150 that handle Docker cleanup scheduling
   • Keep other server management tasks intact

4. Test the implementation

   • Verify hourly execution with test servers
   • Check timezone handling
   • Validate cloud subscription filtering
   • Monitor for duplicate job prevention via WithoutOverlapping middleware

5. Deploy strategy

   • First deploy updated ScheduledJobManager
   • Monitor logs for successful hourly executions
   • Once confirmed, remove cleanup from ServerManagerJob
   • No database migrations required

Implementation Notes

Successfully migrated Docker cleanup scheduling from ServerManagerJob to ScheduledJobManager.

Changes Made:

1. Added processDockerCleanups() method to ScheduledJobManager that processes all servers with a single frozen execution time
2. Implemented getServersForCleanup() to fetch servers with proper cloud/self-hosted filtering
3. Implemented shouldProcessDockerCleanup() for server eligibility validation
4. Removed Docker cleanup logic from ServerManagerJob (lines 136-150)

Key Improvements:

• All servers now evaluated against the same timestamp, ensuring consistent hourly execution
• Proper cloud subscription checks maintained
• Backwards compatible - no database migrations or settings changes required
• Follows the same proven pattern used for database backups

Files Modified:

• app/Jobs/ScheduledJobManager.php: Added Docker cleanup processing
• app/Jobs/ServerManagerJob.php: Removed Docker cleanup logic

Testing:

• Syntax validation passed
• Code formatting verified with Laravel Pint
• PHPStan analysis completed (existing warnings unrelated to changes)