What does this do?

When multiple jobs completed at the same time, NextRun() could return stale/incorrect values due to a race condition in the nextScheduled cleanup logic. The issue occurred because two different methods (selectExecJobsOutCompleted and updateNextScheduled) were cleaning up the nextScheduled slice with different logic:

• One used !t.Before(now) (keeps times >= now)
• One used t.After(now) (keeps times > now)

When these methods ran concurrently on different jobs, their writes could interleave and overwrite each other with inconsistent states.

Solution

Unified both methods to use identical cleanup logic: t.After(now) (keep only strictly future times). This makes the cleanup idempotent - both methods now produce the same result regardless of execution order, eliminating the race condition.

Which issue(s) does this PR fix/relate to?

• fixes issue [BUG] - Job's NextRun time is inaccurate. The job has already run, but NextRun has not been updated to the next scheduled time. #904
• scheduler.go: Updated selectExecJobsOutCompleted and updateNextScheduled to use consistent t.After(now) logic for removing past scheduled times
• job_test.go:
  • Added TestJob_NextRun_MultipleJobsSimultaneously to reproduce and verify the fix for the original bug
  • Added TestJob_NextRun_ConcurrentCompletions to stress-test concurrent NextRun() queries
  • Fixed TestJob_NextRuns weekly job assertion - with proper cleanup, all intervals are now consistently 14 days (no special case for first run needed)

Testing

All existing tests pass. New regression tests verify NextRun() returns correct values when multiple jobs complete simultaneously.

Notes