Running a replication to move from RAIDz2 to RAIDz1, both on TNC 13.0-U6.2, while the documentation provides some information, it seems to lack more details for the process. I also found some youtube videos using the basic replication wizard which provide more details than the docs.
After the replication was successful, I noticed there was a difference in the total size of the data transfered between the two systems. Both are using lz4 compression, but the old system is running RAIDz2 while the new system is running RAIDz1. The total data size on RAIDz2 is 446GB while on RAIDz1 is 436GB. Is there a way to verify data integrity between both systems?
I doubt that the data stored on both systems will ever match 100% unless you have no snapshots being captured, other background tasks running that cause the file system to grow even if the current amount of stored / replicated data remains the same.
For me, the more important measure is whether the snapshots transferred or not. The snapshot contains CRC data to ensure no bit rot, ditto the datasets getting replicated, etc. So as long as you are backing up everything you meant to back up, I’d not worry.
The size difference is absolutely typical of the different levels of raidz and different levels of padding. Send the 436 GB to a raidz3 and you’l get, say, 460 GB; to a mirror, 425 GB.
ZFS is, in itself, your integrity verification process. You can always run a scrub.
Since the replication is a block level transfer, it should be accurate, specially since there were not error reporting during replication. As far as the all the snapshots, they are deleted and recreated every 24 hours for testing.
I don’t think the snapshots have CRC since it is metadata about the data, I know there CRC for the files, but I am unsure if they are including in the snapshot. In other filers that worked with, there was an option to include/exclude the CRC which it was rarely used.
I found some discrepancies in the files sizes due to some files failing to be transferred. I am curious to find out why I didn’t get any alerts regarding the failures though. I am guessing because the rest of the files were successfully transferred? Not sure. Using a comparing tools, it point out the discrepancy in 6 different folders. Of course I had the exclude the NFS exports since the tools only supports SMV/CIFS.
