IBM Support

NDMP data ingestion to IBM Spectrum Protect server storage pools appears to hang

Troubleshooting


Problem

When Network Data Management Protocol (NDMP) is used for backup operations to a primary storage pool, copy storage pool, active-data pool, directory-container storage pool, or cloud-container storage pool on the IBM® Spectrum Protect server, these limitations might occur:

  • NDMP backup operations to container storage pools can take longer to complete than backup operations to primary storage pools, copy storage pools, or active-data pool storage pools.
  • An outstanding database transaction that can pin the Db2 log might occur because of the extended session times on all storage pool types. The larger the NDMP backup image, the more likely it is that a pinned log condition might occur.
  • Restore operations might take a significant amount of more time when data deduplication is enabled on primary storage pools, copy storage pools, and active-data pool storage pools.

Symptom

Db2 log sizes grow exponentially and NDMP sessions to the server appear to hang during backup of data or table-of-contents (TOC) creation.

There might also be messages indicating that the Db2 database restarted while the data ingestion operation is ongoing. The restarts can eventually lead to a failure of the backup operation.

Cause

The problem is caused by a significantly large NDMP backup and the corresponding transaction handling in all storage pool types.

Environment

Any IBM Spectrum Protect server that supports NetApp network-attached storage (NAS) device backup operations to storage pools.

Diagnosing The Problem

During large backups of NDMP data, the log might grow exponentially. If you believe that your system is affected in this manner, collect monitoring data for the server. Then, contact IBM Support to diagnose the problem.

Resolving The Problem

These limitations were fully addressed in the IBM Spectrum Protect server Version 8.1.10.000.

For earlier versions, complete one of the following workarounds:

  • To limit the size of the NetApp volumes that is backed up, divide the volumes that use NDMP directory-level backup.
  • Direct large backups to primary storage pools, copy storage pools, or active-data pool storage pools by setting the MAXSIZE and NEXTSTGPOOL parameters in the storage pools.
  • Use the IBM Spectrum Protect client's automatic snapshot differential ("SnapDiff") backup. Using this type of backup has the following advantages:
  1.        You can use this method to back up data to all types of storage pools.
  2.        There is no limit to the volume size. 
  3.        True, progressive, incremental backups take place.


In addition to the workarounds, it is optimal to take the following actions:

  • Run large NDMP backups when there is less concurrent activity on the server.
  • Disable TOC creation that can significantly reduce the amount of time that a transaction is open.

[{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SSEQVQ","label":"IBM Spectrum Protect"},"ARM Category":[{"code":"a8m50000000Ci5XAAS","label":"Server"}],"ARM Case Number":"","Platform":[{"code":"PF002","label":"AIX"},{"code":"PF016","label":"Linux"},{"code":"PF033","label":"Windows"}],"Version":"8.1.3;8.1.4;8.1.5;8.1.6;8.1.7;8.1.8;8.1.9","Line of Business":{"code":"LOB26","label":"Storage"}}]

Document Information

Modified date:
21 July 2020

UID

ibm10744203