Computer system, control method, and recording medium
In recovery processing, a distributed storage system 100 locks a storage area in which user data for recovery that is user data serving as data for recovery is stored in drives 405 , and releases the lock after caching the user data for recovery in a memory 401 of any of computer nodes 101 . Then, the distributed storage system 100 recovers recovery target data on basis of the user data for recovery cached in the memory 401 . In write processing, the distributed storage system 100 updates the user data for recovery cached in the memory 401 in accordance with a write request and then writes the user data in accordance with the write request.
1. A computer system, comprising a plurality of computer nodes each including a storage device and a memory, wherein:
the storage device of each of the computer nodes is configured to store therein stripe data including user data and a redundant code based on the user data as data elements in a distributed manner;
recovery processing for recovering recovery target data that is any of the data elements on basis of data for recovery that is other data element includes locking a storage area in which user data for recovery that is the user data serving as the data for recovery is stored in the storage device, releasing the lock after the user data for recovery is cached in the memory of any of the computer nodes, and recovering the recovery target data on basis of the user data for recovery cached in the memory; and
write processing in accordance with a write request for the storage area includes updating the user data for recovery cached in the memory in accordance with the write request and writing the user data on the storage area in accordance with the write request.
2. The computer system according to claim 1 , wherein:
the recovery processing is correction read processing in accordance with a correction read request for reading the recovery target data while recovering the recovery target data; and
the recovered recovery target data is transmitted to a transmission source of the correction read request.
3. The computer system according to claim 1 , wherein the recovery processing is rebuilding processing for recovering the user data serving as the recovery target data and writing the user data on the storage device corresponding to the user data.
4. The computer system according to claim 1 , wherein the recovery processing is rebuilding processing for recovering the redundant code serving as the recovery target data and writing the redundant code on the storage device corresponding to the redundant code.
5. The computer system according to claim 1 , wherein:
in the recovery processing, cache management information indicating an area in the memory in which the data for recovery is cached is generated for each hash value based on the data for recovery; and
in the write processing, the data for recovery that is an update target is searched from the memory on basis of the hash value based on the data for recovery and the cache management information.
6. The computer system according to claim 5 , wherein the cache management information indicates the area for each computer node in which the data for recovery is stored.
7. The computer system according to claim 1 , wherein:
one of the computer nodes that has received the write request is configured to transmit, as an intermediate code, a difference between the data for recovery stored in the storage device and the user data written in accordance with the write request to one of the computer nodes having a memory in which the data for recovery is cached; and
the computer node having the memory in which the data for recovery is cached is configured to update the data for recovery cached in the memory on basis of the intermediate code.
8. A control method for a computer system, the computer system comprising a plurality of computer nodes each including a storage device and a memory, the storage device of each of the computer nodes storing therein stripe data including user data and a redundant code based on the user data as data elements in a distributed manner, wherein:
storing, in the storage device of each of the computer nodes, stripe data including user data and a redundant code based on the user data as data elements in a distributed manner;
locking, in recovery processing for recovering recovery target data that is any of the data elements on basis of data for recovery that is other data element, a storage area in which user data for recovery that is the user data serving as the data for recovery is stored in the storage device, releasing the lock after the user data for recovery is cached in the memory of any of the computer nodes, and recovering the recovery target data on basis of the user data for recovery cached in the memory; and
updating, in write processing in accordance with a write request for the storage area, the user data for recovery cached in the memory in accordance with the write request and writing the user data on the storage area in accordance with the write request.