A destination's queue keeps growing

Symptom: the Queue is backing up or Instances are waiting alert, or the Nodes tab shows a growing queue and an old oldest instance for one destination.

1. Is it unreachable? The queue shows unreachable and the last error. Nothing is sent until it can connect: fix the network or the destination. Instances wait on disk, using no attempts, and go when it is back. In a group, they move to another member after the failover delay.
2. Is it paused or limited? Its schedule may be holding it, or its standing limit may be below what arrives. A paused destination does not raise the waiting alert.
3. Is it slow? It accepts instances, just not as fast as they arrive. Options:
   • more Associations / requests per node, if the destination accepts more parallel associations;
   • a larger Asynchronous window, if it supports asynchronous operations;
   • send As received rather than converting (compressing to JPEG 2000 costs CPU on the node);
   • for DICOMweb, more Instances per request.
4. Is it retrying the same instances? Retryable refusals back off up to the Max retry delay. The last error says why; see Dead letters keep appearing for what the statuses mean.

Watch the node’s disk meanwhile: a queue that grows for long enough fills the data volume. See The disk is filling up.