Checking the validity of blocks and backup duplicates of blocks during block reads
View Patent ↗File system coherency of a file system, particularly for system blocks, may be maintained by (i) duplicating and verifying the (system) blocks, (ii) writing (system) blocks as a sequence of atomic commands so that at any given time, there is a valid (system) block on a physical storage device, and/or (iii) defining the file system directory data structure with directory blocks, each of the directory blocks having at least one of a directory entry and a pointer, where updates to the directory data structure are performed atomically.
1. In a distributed file system in which a block of data is stored on a first non-volatile storage and a backup duplicate of the block of data is stored on a second non-volatile storage, a method for reading the block of data from the first non-volatile storage, the method comprising:
a) determining whether the block of data is valid;
b) in response to determining that the block of data is not valid:
i) determining whether the backup duplicate of the block of data is valid, and
ii) in response to determining that the backup duplicate of the block is valid, using the backup duplicate of the block of data to rewrite the block of data to the first non-volatile storage medium; and
c) in response to determining that the block of data is valid, checking the validity of the backup duplicate of the block of data;
wherein the block of data is stored in a first particular part of the first non-volatile storage, and
wherein the validity of the backup duplicate in the second non-volatile storage is determined using a mapping, in a partition control block of the first non-volatile storage, of the first particular part of the first non-volatile storage to a second particular part of the second non-volatile storage for storing backup duplicates of blocks of data stored in the first particular part of the first non-volatile storage.
2. The method of claim 1 wherein the act of checking the validity of the backup duplicate of the block of data is done asynchronously.
3. The method of claim 1 wherein the act of determining whether the block is valid includes checking a checksum value of the block.
4. The method of claim 1 wherein the act of determining whether the block is valid includes comparing a first stamp at a beginning of the block with a second stamp at an end of the block.
5. The method of claim 4 wherein the second stamp is a predetermined function of the first stamp, and
wherein the act of determining whether the block is valid includes determining whether or not the second stamp is the predetermined function of the first stamp.
6. The method of claim 4 wherein at least one of the first and second stamps is a 64-bit number.
7. The method of claim 1 wherein the first non-volatile storage is part of a first file server and wherein the second non-volatile storage is part of a second file server.
8. The method of claim 7 wherein the first file server is remote from the second file server.
9. A machine readable storage medium having machine readable instructions, the machine readable instructions, when executed by a machine, effecting a method for use in a distributed file system in which a block of data is stored on a first non-volatile storage and a backup duplicate of the block of data is stored on a second non-volatile storage, for reading the block of data from the first non-volatile storage, the method comprising:
a) determining whether the block of data is valid;
b) in response to determining that the block of data is not valid:
i) determining whether the backup duplicate of the block of data is valid, and
ii) in response to determining that the backup duplicate of the block is valid, using the backup duplicate of the block of data to rewrite the block of data to the first non-volatile storage medium;
c) in response to determining that the block of data is valid, checking the validity of the backup, duplicate of the block of data;
wherein the block of data is stored in a first particular part of the first non-volatile storage, and
wherein the validity of the backup duplicate in the second non-volatile storage is determined using a mapping, in a partition control block of the first non-volatile storage, of the first particular part of the first non-volatile storage to a second particular part of the second non-volatile storage for storing backup duplicates of blocks of data stored in the first particular part of the first non-volatile storage.
10. The machine readable medium of claim 9 wherein the act of checking the validity of the backup duplicate of the block of data is done asynchronously.
11. The machine readable medium of claim 9 wherein the act of determining whether the block is valid includes checking a checksum value of the block.
12. The machine readable medium of claim 9 wherein the act of determining whether the block is valid includes comparing a first stamp at a beginning of the block with a second stamp at an end of the block.
13. The machine readable medium of claim 12 wherein the second stamp is a predetermined function of the first stamp, and
wherein the act of determining whether the block is valid includes determining whether or not the second stamp is the predetermined function of the first stamp.
14. The machine readable medium of claim 12 wherein at least one of the first and second stamps is a 64-bit number.
15. The machine readable medium of claim 9 wherein the first non-volatile storage is part of a first file server and wherein the second non-volatile storage is part of a second file server.
16. A file server for use in a distributed file system in which a block of data is stored on a first non-volatile storage of the file server, and in which a backup duplicate of the block of data is stored on a second non-volatile storage of another file server, the file server comprising:
a) the first non-volatile storage; and
b) a processor for reading a block of data from the first non-volatile storage, the processor being adapted for
i) determining whether the block of data is valid;
ii) in response to determining that the block of data is not valid:
A) determining that the backup duplicate of the block of data is valid, and
B) in response to determining that the backup duplicate of the block is valid, using the backup duplicate of the block of data to rewrite the block of data to the first non-volatile storage medium;
iii) in response to determining that the block of data is valid, checking the validity of the backup duplicate of the block of data;
wherein the block of data is stored in a first particular part of the first non-volatile storage, and
wherein the validity of the backup duplicate in the second non-volatile storage is determined using a mapping, in a partition control block of the first non-volatile storage, of the first particular part of the first non-volatile storage to a second particular part of the second non-volatile storage for storing backup duplicates of blocks of data stored in the first particular part of the first non-volatile storage.
17. The file server of claim 16 wherein the processor checks the validity of the backup duplicate of the block of data asynchronously.
18. The file server of claim 16 wherein the block of data includes a checksum value, and wherein the act of determining whether the block is valid includes checking the checksum value of the block of data.
19. The file server of claim 16 wherein the block of data includes a first stamp at its beginning and a second stamp at its end, and wherein act of determining whether the block is valid includes comparing the first stamp with the second stamp.
20. The file server of claim 19 wherein the second stamp is a predetermined function of the first stamp, and
wherein the act of determining whether the block is valid includes determining whether or not the second stamp is the predetermined function of the first stamp.
21. The file server of claim 19 wherein at least one of the first and second stamps is a 64-bit number.