Monitoring/detecting failing disks

Currently, OnApp Storage does not check for a failing disk, where the performance is gradually deteriorating.

However, there are plans in the roadmap to alert customers, when there are degraded VDisks on the system (one of the paths has failed), and also when there are disk drives that are failing (hard disk - mechanical failures).

OnApp Storage currently handles I/O errors in the case of replicated storage. It detects and masks "bad sectors" if there is a live replica available. The replica with the I/O errors is taken offline as failed (marking the data store as degraded). If no replicated configuration is used (i.e. a single drive or just a striped data store), OnApp Storage reports the received I/O errors to the filesystem or application level, where recovery should be handled.

OnApp Storage does not currently handle "silent data corruption" on the disk (i.e. requests that do not show up as errors but have "stale/rotten" bits on the drive), but that behavior should be eventually detected and corrected by the application or the filesystem. In the worst case, if corruption is not recoverable, we would need to restore the vdisk from backups.