Repairer raises MustForceRepairError in cases where it shouldn't #874

New Issue

tahoe-lafs · 2009-12-29T21:00:48Z

davidsarah commented

2009-12-29 21:00:48 +00:00

Brian wrote in #786:

I found a bug in the repairer that incorrectly classifies an unrecoverable-but-not-completely-gone file (i.e. one with 1 or 2 shares, when k=3). It raises a MustForceRepairError with an explanation about there being unrecoverable newer versions. The intention was to raise this error when e.g. there are 9 shares of version 1, and 2 shares of version 2, since in that situation, repairing the only recoverable version (1) will knowingly discard the remaining shares of version 2, abandoning any hope of recovering the current contents of the file.

Brian wrote in #786: > I found a bug in the repairer that incorrectly classifies an unrecoverable-but-not-completely-gone file (i.e. one with 1 or 2 shares, when k=3). It raises a `MustForceRepairError` with an explanation about there being unrecoverable newer versions. The intention was to raise this error when e.g. there are 9 shares of version 1, and 2 shares of version 2, since in that situation, repairing the only recoverable version (1) will knowingly discard the remaining shares of version 2, abandoning any hope of recovering the current contents of the file.

tahoe-lafs added the

labels 2009-12-29 21:00:48 +00:00

tahoe-lafs added this to the undecided milestone 2009-12-29 21:00:48 +00:00

warner commented

2009-12-29 23:43:31 +00:00

changeset:ba0690c9d7a3bc28 fixes this, and returns results with successful=False when there are insufficient shares for repair and there is only one version visible.