Symptoms

Data Domain destination system configured with MTREE replication may report higher usage if the source system has fastcopy files in the same MTREE.

Cause

By default fastcopy operation on Data Domain file system (ddfs) does not set the Base File tracking (BFT) flag. During MTREE replication, the fastcopied file is replicated as a gen0 because, during ingest, fastcopy of the directory did not have the BFT flag turned on. Also, fastcopy files shares L0 segments with the original file, hence it does require additional space. Due to this, it may show a higher compression factor on the source system compare to the destination system. Example: Information from sfs_dump from the source and destination Data Domain system SRC fast copy file: /data/col1/MTREE1/increment/backdata1: mtime: 1568680103968447000 fileid: 3751127 size: 5893005312 type: 9 seg_bytes: 5913233056 seg_count: 708373 redun_seg_count: 708373 (100%) pre_lc_size: 0 post_lc_size: 0 (0%) mode: 02000100640 start_offset: 0 end_offset: 18446744073709551615 DST: /data/col1/MTREE1/increment/backdata1: mtime: 1568680103968447000 fileid: 3751127 size: 5893005312 type: 9 seg_bytes: 5912862460 seg_count: 708308 redun_seg_count: 704269 (99%) pre_lc_size: 41811242 post_lc_size: 27638454 (66%) mode: 02000100640 start_offset: 0 end_offset: 18446744073709551615 In the above example, since BFT is not set, hence it was replicated destination as gen0 file showing post_lc_size value.

Resolution

Turn on BFT for fastcopy using synthetic-optim option. Note that this option is not documented. #filesys fastcopy source <src dir> destination <dst dir> synthetic-optim Note: This setting is applied to newer fasctcopy files. It is not possible to reclaim data for data replicated without BFT. User must wait for fast-copy files to expire or a manual clean-up is required.

Support Cases