X_JOB_SLOT_LOST
Means
the fleet concurrency slot this run held was taken by another worker
Typical cause
this worker stalled past the slot TTL and a second worker renewed slot 0 for itself, so job.concurrency was about to be exceeded. Its own code and not X_JOB_LEASE_LOST because the queue may still consider this worker the owner of the job — the two are different rows on different clocks (x_job_leases vs x_jobs.visible_at), and only one of them was lost. The run is aborted rather than left to double up
Fix
x jobs show <id> --json — a run that keeps losing its slot is slower than its slot TTL, so raise the TTL or make the job shorter