@kapa.ai What do these errors from the workload-api-server component mean?

[2026-03-19  13:12:36]	2026-03-19 03:12:36,327 [io-executor-thread-6]	ERROR	i.a.w.s.WorkloadService(launchWorkload):223 - Failed to update workload (351945a3-3a0e-4b20-a617-72505a691404_5535_0_check) to launched. Current status is launched.
[2026-03-19  13:12:42]	2026-03-19 03:12:42,806 [io-executor-thread-4]	ERROR	i.a.w.s.WorkloadService(runningWorkload):274 - Failed to update workload (351945a3-3a0e-4b20-a617-72505a691404_5535_0_check) to running. Current status is running.

These errors from the

WorkloadService

indicate a workload lifecycle state conflict — the system is attempting to transition a workload to a state it is already in: •

Failed to update workload (...) to launched. Current status is launched.

— The launcher is trying to mark a workload as "launched," but it is already in the "launched" state in the database. •

Failed to update workload (...) to running. Current status is running.

— Similarly, the system is trying to mark the workload as "running," but it is already recorded as "running." This suggests the workload lifecycle is not progressing to a terminal state (e.g., success or failure). As a result, the workload stays stuck in an intermediate state ("launched" or "running") indefinitely. [Jobs not removed when completed] A known root cause for this pattern is a deadlock in the workload launcher: 1. Workloads complete (pods finish running), but their status in the database is never updated from "launched"/"running" to a terminal state like "success." 2. The launcher tries to claim these workloads again, but the claim API rejects them because they are already in a non-pending state. 3. Because the claim fails, the

CHECK_STATUS

stage is skipped, so the launcher never detects that the pod has completed and never updates the status. 4. This creates a catch-22: workloads stay stuck, claims keep failing, and the system jams. [Airbyte services stuck] In practice, this can manifest as K8s job pods never being cleaned up after completion. [Jobs not removed when completed] If you are seeing these errors, it is worth checking: • Whether the corresponding K8s pods have actually completed but their workload status remains stuck. • Whether there are any connectivity or database issues preventing status updates from being persisted. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.