Dropped write error detection
AssigneeInternational Business Machines Corporation
InventorsGang Lyu, Jun Gu, Hong Zhou
Life of this patent
Estimating the term…
Abstract
A computer-implemented method for dropped write error detection is proposed. In the method, a read request for a stride stored in an array of storage drives is received. The stride includes segments of a data and a first parity associated with the data spreading across the storage drives in the array of the storage drives. In response to the read request being a predefined sequential read request and a state of the stride being a first state, a parity check is performed on the stride. The first state indicates that no parity check has been performed after the data is written into the array of storage drives. The state of the stride is changed to a second state, and the second state is different with the first state.
Claims18
A computer-implemented method, comprising:
- receiving, by one or more processing units, a read request for a stride stored in an array of storage drives, wherein the stride includes segments of a data and a first parity associated with the data spreading across the storage drives in the array of the storage drives;
- in response to the read request being a predefined sequential read request and a state of the stride being a first state, performing, by one or more processing units, a parity check on the stride, wherein the first state indicating that no parity check has been performed after the data is written into the array of storage drives;
- and
- changing, by one or more processing units, the state of the stride to a second state, the second state being different with the first state.
A computer program product (CPP) comprising a computer readable storage medium having program instructions embodied therewith, the program instructions being executable by a device to perform a method comprising:
- receiving a read request for a stride stored in an array of storage drives, wherein the stride includes segments of a data and a first parity associated with the data spreading across the storage drives in the array of the storage drives;
- in response to the read request being a predefined sequential read request and a state of the stride being a first state, performing a parity check on the stride, wherein the first state indicating that no parity check has been performed after the data is written into the array of storage drives;
- and
- changing the state of the stride to a second state, the second state being different with the first state.
A computer implemented method (CIM) for use with a plurality of cooperatively operating storage devices that collectively store a plurality of strides of data, the CIM comprising:
- writing a first stride of the plurality of strides to the plurality of cooperatively operating storage devices, with the first stride including substantive data and metadata including information indicating that no parity check has been performed since the substantive data was written to the plurality of cooperatively operating storage devices;
- receiving an access request to access the substantive data of the first stride;
- determining that the first stride has not been subject to a parity check since the first stride was written to the plurality of cooperatively operating storage devices based on the metadata of the first stride;
- and
- responsive to the access request and further responsive to the determination that the first stride has not been subject to a parity check:
- performing a parity check on the first stride to determine that the first stride has been stored correctly,
- responsive to the determination that the substantive data of the first stride has been stored correctly and further responsive to the receipt of the access request, providing access to the substantive data of the first stride, and
- changing the metadata of the first stride to indicate that a parity check has been performed on the first stride.
Dependencies are inferred from the claim text.