Comment 9 for bug 633101

I get this occasionally over an SFTP interface to a remote block storage system from time-to-time. Due to their implementation there is an occasional and variable but short delay in newly written files becoming accessible in subsequent requests. As its a question of timing it usually doesn't cause a problem, but sometimes it does and then duplicity is unable to perform subsequent backups without manually wiping the local cache.

Note the first run terminated with the error `Giving up after 1 attempts. IOError: [Errno 2] No such file`. Followed by the assertion errors described above for all subsequent runs until the cache is cleared.

We may take the view that it is reasonable for the program to give up when during the original error on the remote, but I think it should to handle corruption in its own cache without manual intervention.