Method and an apparatus for aggressively detecting media errors on storage devices with negligible performance impact
A method and an apparatus to aggressively detect media errors on storage devices with negligible performance impact have been disclosed. In one embodiment, the method includes computing a first data rate at or above which a scrub on a set of storage devices can be completed within a predetermined period of time, and running the scrub on the storage devices at a second data rate without interfering with the storage devices servicing client requests to access the storage devices, the second data rate being at or above the first data rate.
1. A computer-readable storage medium that stores instructions which, if executed by a processor, will cause the processor to perform operations comprising:
causing a plurality of storage devices to perform a scrub to check for media error without interfering with the plurality of storage devices servicing client requests; and
dynamically changing a rate at which the scrub is performed in response to a volume of the client requests.
2. The computer-readable storage medium of claim 1 , wherein causing the plurality of storage devices to perform the scrub comprises issuing verify requests to the plurality of storage devices.
3. The computer-readable storage medium of claim 1 , wherein the verify requests include Small Computer System Interface (SCSI) verify requests.
4. The computer-readable storage medium of claim 1 , wherein the operations further comprise:
initiating a parity scrub on one of the storage devices; and
suspending the scrub in response to an initiation of the parity scrub.
5. The computer-readable storage medium of claim 4 , wherein the operations further comprise:
resuming the scrub after the parity scrub has stopped.
6. A method comprising:
computing a first data rate at or above which a plurality of storage devices can be completely scrubbed for media errors within a predetermined period of time;
causing the plurality of storage devices to perform a scrub to detect media errors at a second data rate without interfering with the plurality of storage devices servicing client requests to access the plurality of storage devices, the second data rate being at or above the first data rate computed, wherein causing the plurality of storage devices to perform the scrub comprises issuing verify requests from a storage server to the plurality of storage devices to cause the plurality of storage devices to check whether data in the plurality of storage devices can be read correctly without transferring the data from the plurality of storage devices to the storage server, the verify requests include Small Computer System Interface (SCSI) verify requests;
determining a volume of the client requests;
dynamically adjusting the second data rate in response to the volume of the client requests; and
repairing a storage device if a media error is detected on the storage device during the scrub.