Hello.

I was evaluating ScaleIO for some time. It was quite stable, but I decided to try how reliable it really is, and was performing some cluster modifications to see how it will perform.

So, I added additional SDS (had 3, added 1). Fine

Added new SSD disk and include it into SSD pool.

Immediately after that action I started to get different IO errors on XFS filesystem and finally I have got "xfs log i/o error detected. shutting down filesystem". The server with SDC was nearly idle, so it was probably doing 1-5 IOPS.

I decided to save my data by unplugging the new disk. I can't say that this helped, because the filesystem was already shut down. I unmounted, checked and remounted filesystem and everything ended ok.

What was that? Could ScaleIO read data from uninitialized disk and provide it as valid? Do it has any checksums to verify the data? Is data corruption possible not due to multiply hardware failure?