List of corruption cases for scrub
Hi.
I tried to a create list of corruption test scenarios for scrubbing process with RAID5. Here's the list: https://btrfs.wiki.kernel.org/index.php/Scrub_corruption_cases
If similar list already exists in btrfs wiki, let me know, I'll move this data to there. Feel free to add/update/delete entries or suggest new corrupt ideas!
Cheers.
Lakshmipathi.G
Hi,
Quoted 7 lines · Lakshmipathi.G
Hi.
I tried to a create list of corruption test scenarios for scrubbing process with RAID5. Here's the list: https://btrfs.wiki.kernel.org/index.php/Scrub_corruption_cases
If similar list already exists in btrfs wiki, let me know, I'll move this data to there. Feel free to add/update/delete entries or suggest new corrupt ideas!
Thanks for doing that. About the test "Device-corruption", I suppose that you want to test a scenario where one disk fails. If so you do not have to always corrupt "D0" block, but the first block of the file which is stored in the failing disk (which could be different from the D0).
Another scenario which should be tested, is the reshaping of the raid: add a drive (or remove a drive) then re-balance the system (with or without some corruption). IIRC this use the same scrub code.
Quoted 4 lines
Cheers.
Lakshmipathi.G
--
BR
G.Baroncelli
Signature · 2 lines
gpg @keyserver.linux.it: Goffredo Baroncelli <kreijackATinwind.it> Key fingerprint BBF5 1610 0B64 DAC6 5F7D 17B2 0EDA 9B37 8B82 E0B5
Quoted 4 lines · Goffredo Baroncelli
Hi,
Quoted 2 lines · Lakshmipathi.G
Hi.
Quoted 8 lines
Quoted 5 lines
I tried to a create list of corruption test scenarios for scrubbing process with RAID5. Here's the list: https://btrfs.wiki.kernel.org/index.php/Scrub_corruption_cases
If similar list already exists in btrfs wiki, let me know, I'll move this data to there. Feel free to add/update/delete entries or suggest new corrupt ideas!
Thanks for doing that. About the test "Device-corruption", I suppose that you want to test a scenario where one disk fails. If so you do not have to always corrupt "D0" block, but the first block of the file which is stored in the failing disk (which could be different from the D0).
Yes, that's right. Updated (Device-corruption) the list now.
Quoted 2 lines
Another scenario which should be tested, is the reshaping of the raid: add a drive (or remove a drive) then re-balance the system (with or without some corruption). IIRC this use the same scrub code.
Added this as 'Rebalance-corruption' on the wiki. thanks!
Cheers.
Lakshmipathi.G
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 8 lines · Qu Wenruo
Quoted 5 lines · Lakshmipathi.G
Hi.
I tried to a create list of corruption test scenarios for scrubbing process with RAID5. Here's the list: https://btrfs.wiki.kernel.org/index.php/Scrub_corruption_cases
Quoted 7 lines
Quoted 2 lines
If similar list already exists in btrfs wiki, let me know, I'll move this data to there. Feel free to add/update/delete entries or suggest new corrupt ideas!
What about adding testing result for latest kernel release?
Thanks,
Qu
True, Ideally, the wiki should have list-corruption cases along with result for each latest kernel version.
Planning to add more cases, gradually. For example, I think inode object with RAID5 can be replicated across drives. If thats correct, can scrub fix those corruption too? cases like this may be more important than other to-do items like raid5 with features like snapshot/quota etc
I think it should be possible to add test-results for few cases but list has repeated-cases with different sizes,drives etc for these cases we need some kind of automated scripts.
Hope to extend the cases first and then start updating results with manual or automated tests.
Cheers.
Lakshmipathi.G