You have configured a Data Domain Storage Unit as a Veeam backup repository using MTree replication. Data Domain replication results in PANIC on the target Data Domain storage.
Veeam Backup & Replication leverages the Virtual Synthetic (VS) feature of the DD Boost library for building Grandfather-Father-Son (GFS) restore points or when applying retention policies to backup chains. As part of the Veeam workflow, overwrites can occur over sections of the file being replicated that have already been processed. This behavior will trigger the recipe tracking process to remove all recipes on the disk, but not any invalid in-memory recipes. This may result in a corrupted recipe to be recalled and lead to an incorrect recipe replication process later (because of a checksum mismatch) so that recipe replication will be confused and target EMC Data Domain may PANIC.
Turning off the recipe processing feature on the source Data Domain system source for MTree replication will resolve the issue. A later DDOS 6.x release will fix the underlying issue, and recipe processing can then be re-enabled for performance reasons.
Customers should first verify their system matches the criteria for overcoming the issue:
1. Confirm that the system configured for VEEAM backups is also configured for Data Domain MTree replication:
# replication show config
In the output look for "mtree://"
2. Confirm if managed file replication (MFR) or backup replication via DD Boost) is being used, on the Data Domain you can use this command:
# ddboost file-replication show active all
Also verify in the backup application configurations. Once confirmed through the commands above your system may be a configuration where this issue applies, please do following or contact Dell/EMC support for assistance:
3. Disable running replication contexts and MFR tasks, if any.
4. Run command
#se sysparam set RECIPE_REPL_ENABLED=FALSE
5. Enable replication contexts and resume MFR tasks, if any.
Note the workaround to be performed by a Dell EMC TSE will require disabling Data Domain replication and MFR replication jobs, and will require downtime for the workaround to be completed.