build a checker/verifier that can work from just SI #482

Open
opened 2008-07-01 23:41:08 +00:00 by warner · 2 comments

The current checker code requires a read-cap, or at least a verifier cap. We should have a checker/verifier that can work from just the storage index. It should read all shares and verify all hashes against the UEB. It should compare the UEB from each share against each other and reject the minority versions. If a verifier cap is available, it should compare the UEB from the shares against the UEB hash from the verifier cap.

Files which fail the check (fewer shares than expected) should be reported. Shares which fail their hash checks (or disagree with the majority, or with the verifier cap) should be reported.

The current checker code requires a read-cap, or at least a verifier cap. We should have a checker/verifier that can work from just the storage index. It should read all shares and verify all hashes against the UEB. It should compare the UEB from each share against each other and reject the minority versions. If a verifier cap is available, it should compare the UEB from the shares against the UEB hash from the verifier cap. Files which fail the check (fewer shares than expected) should be reported. Shares which fail their hash checks (or disagree with the majority, or with the verifier cap) should be reported.
warner added the
code-encoding
major
task
1.1.0
labels 2008-07-01 23:41:08 +00:00
warner added this to the undecided milestone 2008-07-01 23:41:08 +00:00
tahoe-lafs added
enhancement
and removed
task
labels 2009-12-06 00:06:31 +00:00
davidsarah commented 2009-12-06 00:18:57 +00:00
Owner

See also #847.

See also #847.
davidsarah commented 2010-02-11 01:14:05 +00:00
Owner

Would we really need this if we had fixes for #568 and #625? Verification just from the SI would not be able to detect malicious corruption -- only some cases of accidental corruption. Under what circumstances would we have the SI and not the verify cap?

Would we really need this if we had fixes for #568 and #625? Verification just from the SI would not be able to detect malicious corruption -- only some cases of accidental corruption. Under what circumstances would we have the SI and not the verify cap?
Sign in to join this conversation.
No Milestone
No Assignees
2 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Reference: tahoe-lafs/trac-2024-07-25#482
No description provided.