Comparison
SchedulervsUnschedulable workload
Scheduler
you asked for the workload to run somewhere, and something scored every node and picked one before any container existed.
The control-plane component that assigns pending workloads to nodes, filtering out the impossible ones and scoring the rest. It is the origin of a large fraction of platform confusion, because 'the deploy is stuck' usually means a scheduler quietly deciding nothing satisfies your constraints. Its decisions are one-shot at placement time, which is why an already-running workload does not move when the cluster's shape changes.
Full entry →Unschedulable workload
the deploy has been sitting at pending for twenty minutes with no error, because no node satisfies every constraint you set.
A workload the scheduler cannot place, left waiting rather than failed. The scheduler's own event explains why in one line — insufficient memory, no matching label, untolerated taint, unsatisfiable anti-affinity — and reading it is invariably faster than reasoning about it. It is the most common form of 'the deploy is stuck', and it looks like nothing is happening because, precisely, nothing is.
Full entry →