CONTAP-468748: FlexGroup write placement continues to select full constituents even when other constituents have available capacity
Issue
- After a FlexGroup is expanded (additional constituents added), new write and object allocations can continue to prefer older, highly utilized constituents.
- Those older constituents may remain near full (for example ~99% used with 0 B available) even when newer constituents, or other aggregates in the FlexGroup, have substantial free space.
- Client applications writing to the FlexGroup — including ONTAP S3 buckets backed by a FlexGroup, and FabricPool capacity-tier PUT activity directed at that object store — can fail with no-space / object-store-full style errors despite free capacity elsewhere in the FlexGroup.
- On diagnostic review of FlexGroup ingest heuristics for a fully urgent constituent, Urgency can show 100% while Target% remains non-zero (for example around 11%), so the full constituent continues to receive a share of new allocations.
- Moving or resizing constituents can restore usable free space; upgrading alone does not add free space to constituents that are already full, and does not redistribute existing objects.
