Hey, Looks like issues with the Datanodes. The block placement policy isn’t able to get 3 datanodes with available DISK space. Give a check to your Cluster datanodes state, types of storages, space in DISK and related stuff.
Try enabling debug logs, you might see logs like Datanode X is not chosen since <reason> If that is a prod cluster and too busy, the datanodes may be getting excluded with NODE_TOO_BUSY. If so try playing with consider load configs like dfs.namenode.redundancy.considerLoad or dfs.namenode.redundancy.considerLoad.factor Or if some other reason for exclusion, you need to fix that accordingly… -Ayush > On 13-Dec-2021, at 7:43 PM, Claude M <[email protected]> wrote: > replicationFallbacks=[ARCHIVE]} > > What does this mean and how can it be resolved? > > Also
