[tahoe-dev] UnhappinessError during renew

[Brian Warner]

On 1/5/11 6:26 AM, slush wrote:
> Hello,
> 
> after few weeks I checked logs of my storage repairs and found, that
> process is permanently throwing UnhappinessError. That means I probably
> lost some of my data, right? Is here some way how to fix it / skip error
> and let repairer to renew other files of my storage?


By the way, to be precise about our terminology, this sounds like a
failure of *repair*, not a failure of *renew*. "repair" is the process
of creating new shares to replace ones that have been lost (either
because storage servers deleted them, or because storage servers are
unavailable). "renew" means "lease renew", which is the process of
telling the storage server that you want to keep those shares around for
a while longer.

The "--repair" argument to "tahoe deep-check" is what triggers repair.
The "--add-lease" argument is what triggers lease-renewal.

Incidentally, an UnhappinessError during repair doesn't mean you've lost
data. The repairer (at least, until it gets fixed) is being too picky,
and is throwing an error when it cannot make the file as healthy as it
would like (even though it may be perfectly readable). And the #755
error is causing the rest of the deep-traversal to stop when it hits the
first error of any sort.

(if the error said something like Unrecoverable or Unrepairable, then
that might indicate data loss, but UnhappinessError is pretty mild)

cheers,
 -Brian