I have to investigate a recent increase in failing RAID-1 (Mirror) arrays we're encountering in our installed PC park (10K+ PC's).
The problem is twofold: one part of being a pilot phase we started with Solid-State-Drives (SSD's). We rolled out about 80 PC's with 2x Kingston V300 60GB SSD's in RAID-1. On these installs we see a 10-15% failure rate over a period of max. 3 months, which is off course unacceptable. This is on a PC platform with an on-board Intel ICH7R SATA RAID controller
The 2nd part is that we're also seeing an alarming increase in failures on an existing platform with recent Seagate Barracuda HDD's. These systems are based on old P4 platforms and are using the Promise FastTrak TX2300 SATA RAID controller. We've been using these systems with the same RAID controllers for over 7 years, but it's only with the Seagate HDD's used recently that we're seeing an increasing failure rate.
Now on to the problem: What I have with both of these cases, is that I can't determine what caused this RAID array to fail. I received examples of failed systems for both cases. Problem is that when investigating the drives, both drives appear to work fine, report no SMART errors, and have all their files readable (except off course the files created after the array failure).
When erasing the disks and re-creating the RAID array, the system works perfectly in the lab.
Are there any tools I could use to further analyse these disks? I could believe they're working fine most of the time, but due to a short hickup are marked as 'bad' by the RAID controller. It could also be the RAID controller which is faulty I assume? How to test this?
Both mentioned systems run on Windows POSReady 2009 (Basically windows XP)
dd
and the array started magically working again. Same thing happened again a few months later. Switched to a proper controller afterwards and never had any more issues with the same drives. Others will tell you they've used ICHR RAIDs for years without issues but I would never use it again.