Re: ftp.br MD5Sum mismatch
>>>>> "CC" == Carlos Carvalho <carlos@fisica.ufpr.br> writes:
CC> We were finishing updating when you fetched these files. The first
CC> phase ended at 15:05:20 and the second one at 15:05:58. You fetched
CC> InRelease at 15:05:55. You had error for
CC> experimental/.../Translation-en, which you fetched at 15:05:57, and
CC> the other two which you fetched at 15:05:59, after our second phase
CC> had finished. Remeber that rsync updates all files rapidly, after
CC> having downloaded all of them. Therefore it may well be that you
CC> fetched the two InRelease's before they were updated and the other
CC> indices after rsync changed them.
CC> It seems a case of bad luck.
It seems you are saying that with the current mirror software, there is
a guaranteed (even just a) few seconds a day when this error will occur.
That means the software should be improved yet further to eliminate this
error completely!
I mean how do banks or any big company do things? How could there be
even a few seconds a day when this could happen.
So we found a weak point in the algorithm?
Reply to: