Managing uncorrectable user data
A technique for managing user data in a storage system includes accessing RAID metadata to identify user data that the storage system backs with broken RAID arrays. The technique further includes marking metadata that points to at least some of that user data to identify such user data as uncorrectable.
1. A method of managing user data in a storage system, the method comprising:
accessing RAID metadata that identifies a set of broken RAID arrays in which data loss has occurred;
identifying, based on the RAID metadata, multiple pointer structures that point to user data in the set of broken RAID arrays, the pointer structures participating in mapping a set of data objects hosted by the storage system to underlying storage; and
marking at least some of the multiple pointer structures to identify user data that is uncorrectable.
2. The method of claim 1 , wherein the acts of accessing, identifying, and marking are performed by a consistency-checking facility that runs in the storage system, the consistency-checking facility also checking for uncorrectable user data that arises from damaged metadata.
3. The method of claim 2 , wherein identifying the pointer structures that point to the user data in the set of broken RAID arrays includes:
identifying the broken RAID arrays from the RAID metadata;
identifying multiple storage extents within the broken RAID arrays;
accessing descriptors of the multiple storage extents; and
following back-pointers associated with the descriptors to locate the pointer structures.
4. The method of claim 2 , further comprising generating results of the consistency-checking facility, the results identifying one or more data objects mapped to the set of broken RAID arrays and associated addresses at which user data is uncorrectable.
5. The method of claim 1 , wherein the pointer structures are part of a virtual block layer used for mapping the set of data objects.
6. The method of claim 1 , wherein marking at least some of the multiple pointer structures includes marking as uncorrectable pointer structures that point to unreadable user data but not marking as uncorrectable pointer structures that point to readable user data.
7. The method of claim 6 , further comprising distinguishing the readable user data from the unreadable user data based on issuing read requests to the user data pointed to by the multiple pointer structures and determining whether the read requests succeed or fail.
8. The method of claim 7 , further comprising copying the readable user data to a functioning RAID array in the storage system.
9. The method of claim 8 , wherein the broken RAID array is composed of storage strips derived from respective storage drives, the storage drives including a set of working drives and a set of broken drives, and wherein the method further comprises, after copying the readable user data to the functioning RAID array, unmapping the broken RAID array and making the storage strips of the set of working drives available for use in forming new RAID arrays.
10. A computerized apparatus, comprising control circuitry that includes a set of processing units coupled to memory, the control circuitry constructed and arranged to:
access RAID metadata that identifies a set of broken RAID arrays in which data loss has occurred;
identify, based on the RAID metadata, multiple pointer structures that point to user data in the set of broken RAID arrays, the pointer structures forming part of system metadata used for mapping a set of data objects hosted by the computerized apparatus to underlying storage; and
for at least some of the multiple pointer structures, marking the system metadata to identify user data that is uncorrectable.
11. The computerized apparatus of claim 10 , wherein the control circuitry configured to mark at least some of the multiple pointer structures is further configured to mark as uncorrectable pointer structures that point to unreadable user data but not to mark as uncorrectable pointer structures that point to readable user data.
12. The computerized apparatus of claim 11 , wherein the control circuitry is further configured to copy the readable user data to a functioning RAID array.
13. A computer program product including a set of non-transitory, computer-readable media having instructions which, when executed by control circuitry of a computerized apparatus, cause the computerized apparatus to perform a method of managing user data, the method comprising:
accessing RAID metadata that identifies a set of broken RAID arrays in which data loss has occurred;
identifying, based on the RAID metadata, multiple pointer structures that point to user data in the set of broken RAID arrays, the pointer structures participating in mapping a set of data objects hosted by the storage system to underlying storage; and
marking at least some of the multiple pointer structures to identify user data that is uncorrectable.
14. The computer program product of claim 13 , wherein the acts of accessing, identifying, and marking are performed by a consistency-checking facility that runs in the storage system, the consistency-checking facility also checking for uncorrectable user data that arises from damaged metadata.
15. The computer program product of claim 14 , wherein the method further comprises generating results of the consistency-checking facility, the results identifying one or more data objects mapped to the set of broken RAID arrays and associated addresses at which user data is uncorrectable.
16. The computer program product of claim 13 , wherein marking at least some of the multiple pointer structures includes marking as uncorrectable pointer structures that point to unreadable user data but not marking as uncorrectable pointer structures that point to readable user data.
17. The computer program product of claim 16 , further comprising distinguishing the readable user data from the unreadable user data based on issuing read requests to the user data pointed to by the multiple pointer structures and determining whether the read requests succeed or fail.
18. The computer program product of claim 17 , wherein the method further comprises copying the readable user data to a functioning RAID array in the storage system.
19. The method of claim 18 , wherein the broken RAID array is composed of storage strips derived from respective storage drives, the storage drives including a set of working drives and a set of broken drives, and wherein the method further comprises, after copying the readable user data to the functioning RAID array, unmapping the broken RAID array and making the storage strips of the set of working drives available for use in forming new RAID arrays.
20. The computer program product of claim 13 , wherein identifying the pointer structures that point to the user data in the set of broken RAID arrays includes:
identifying the broken RAID arrays from the RAID metadata;
identifying multiple storage extents within the broken RAID arrays;
accessing descriptors of the multiple storage extents; and
following back-pointers associated with the descriptors to locate the pointer structures.