Client data storage or node replication sessions writing to multiple container storage pools that include at least one S3 or Azure cloud-container storage pool can result in unrecoverable data (IT24321)

Content

WHO IS AFFECTED:
All users of the IBM Spectrum Protect Version 8.1 or later server in environments where client data storage or node replication sessions write data to multiple container storage pools, where at least one storage pool is an S3 or Azure cloud-container storage pool, and the cloud-container storage pool has an assigned storage pool directory, are affected. See the PROBLEM SUMMARY section for additional details about the affected storage pool configurations.

WHO IS NOT AFFECTED:
You are not affected by the problem that is described in this document if any of the following conditions are true for your system environment:

• The IBM Spectrum Protect server is earlier than V8.1.
• The environment does not include S3 or Azure type cloud-container storage pools.
• The environment includes S3 or Azure type cloud-container storage pools; however, none of these storage pools has an assigned storage pool directory. In this configuration, the cloud-container storage pool is configured exclusively as a target of a storage tiering rule.
• The IBM Spectrum Protect server has only a single storage pool that is an S3 or Azure type cloud-container storage pool with assigned storage pool directories and that is the only storage pool on the server.

PROBLEM SUMMARY:
In certain server configurations, the problem that is described in APAR IT24321 might result in loss of data.  The following configurations are affected:

Affected client data storage sessions
This problem might occur during client data storage operations when a client uses an INCLUDE option to assign a subset of files to a management class that is different from the default management class defined in the policy for this node. If the management classes direct data to different container storage pools and one of those storage pools is an S3 or Azure cloud-container storage pool with a storage pool directory, this problem might occur.

Affected node replication operations
The destination storage pool that is selected on the target server during node replication operations is determined by the management class that is assigned to each file that is replicated. This problem might occur if the policy definitions on the target replication server are set to write data to different container storage pools, and one of those storage pools is an S3 or Azure cloud-container storage pool with a storage pool directory.

RECOMMENDATION:
Apply the fix for APAR IT24321.

PROBLEM RESOLUTION:
This problem will be resolved in the following IBM Spectrum Protect server levels with APAR IT24321. This is subject to change at the sole discretion of IBM:
     8.1.1.200
     8.1.5.100
     8.1.6

If you require a fix before these levels become available, contact IBM Software Support.

IDENTIFYING AFFECTED DATA:
To determine whether you are affected by the problem that is described in this document, run the following commands by using the same instance user ID that is used to start the IBM Spectrum Protect server:

1. If the database manager is not active, start it:

db2 start database manager

2. Open a connection to the server database. If the server database is not the default, tsmdb1, specify the appropriate database alias for your database:

db2 connect to tsmdb1

Tip: To identify the database name, run the following command:

db2 list database directory

3. Set the default database schema to tsmdb1:

db2 set schema tsmdb1

4. To identify any containers that might be affected by the problem that is described in this document, run the following IBM Db2 SQL command:

db2 "select distinct sdcn.cntrname from sd_all_chunks sdcl inner join sd_containers sdcn on sdcl.cntrid=sdcn.cntrid and sdcl.poolid!=sdcn.poolid where sdcn.type=3 and ( sdcn.cntrname like '%.ncf' or sdcn.cntrname like '%.dcf') "

5. If the output of the Db2 command includes container names, you might be affected by the problem that is described in this document and continue to step 6.

6. Apply the appropriate fix as described in the PROBLEM RESOLUTION section.

7. Conduct an audit of the containers that were displayed by the SQL command in step 4:

AUDIT CONTAINER container_name 

8. To identify the client nodes with damaged data, run the QUERY DAMAGED TYPE=NODE command on all directory and cloud-container storage pools.

For additional assistance with repairing or replacing affected files contact IBM Software Support.