Erasure Coded Data shows 0 bytes for new object store after StorageGRID expansion
Applies to
- NetApp StorageGRID 11.x
- Erasure Coded (EC)
- Information lifecycle management (ILM)
- Newly added or expanded (eg new shelf) node
Issue
After expanding a StorageGRID system with new Storage Nodes, it would appear that Erasure Coded objects are not being ingested to the newly expanded nodes after expansion.
This may give the appearance that there is something wrong with the new Storage Node, and in order to have the new storage nodes accept new EC data an administrator may invoke such actions as the following in order to get the newly expanded nodes to accept new content...
- Modifying Storage Pools
- Modifying Storage Grades
- Modifying Erasure Coding Schemes
- Modifying ILM rules and policies
This may in turn create new issues such as...
- Introduce a backend ILM driven movement of object data to satisfy a new ILM rule or move content to a new pool
- Generate sporadic alerts/alarms in relation to Resource Use
- Client side performance issues due to ILM driven backend workload
- Grid side performance issues due to ILM driven backend workload
- Grid Node Instability and/or Grid Node service instability (IE, Blue Nodes, Services randomly going Up/Down)
In 11.4 and previous releases
- Internal Erasure Coding groups are maintained on existing storage nodes
- New EC groups are created when the existing EC group nears exhaustion, thus automatically creating a new EC group
- The new EC Group inherits the new Storage Nodes
To verify that newly added Object Store shows EC data as 0 bytes even after data is ingested as per ILM policy:
NMS/GMI --> Nodes --> "Newly Expanded Storage Node" --> Storage --> Scroll Down to "Object Stores"
Note that Replicated Data is not zero, so ingests are working.