Our thinking

Data Corruption on Promise Pegasus R4 Thunderbolt RAID

I recently set up a Mac mini Server with a Promise Pegasus R4 RAID.

I reinitialised the array a RAID6. I left the stripe size at the default of 128k, and changed the block size to 4k as this should increase performance.
As the array was initialising, if I copied data to the RAID using the finder, the data would get corrupted. Once the corrupt data was written, every time I went to read it back, I got the same result (verified with md5 hashes)
If I copied data to the RAID using the command-line, it was OK.
If I used the Finder to duplicate a file on the RAID – corruption.
If I copied a file to the RAID over the LAN it seemed a bit better, but occasionally got corruption.
I turned off the write back cache and this didn’t make any difference.
I erased the array and re-initialised it with a block size of 512 bytes and it all (so far) seems OK.
I didn’t wait the 24 hours it seems like it was going to take to see if the array was stable once it had completed initialisation, it’s possible that it would be OK, but I didn’t have time to take a chance on this…

1 thought on “Data Corruption on Promise Pegasus R4 Thunderbolt RAID

  1. The latest Firmware update for the Pegasus series claims to have —

    Fixed data inconsistency error seen on OS X 10.8.2 or later when using logical drives with sector sizes greater than 512 bytes

Leave a Reply