Thinking of which, I think it is a bug for glusterfs to report the size of the bigger disk, it should be reporting the smallest size available.
Thinking yet again, I think the current behavior is better. Reporting the smallest size would be appropriate _only_ if all files are being replicated with equal count. in situations where some files are replicated 2 times, some 3 times and some with no extra replication, the total 'size' cannot be deterministically calculated without actually filling the servers with different types of files. thanks, avati -- Anand V. Avati
