if a device is in drain mode without rebalance running, and a fid with a copy
on it has a replication event, replication may over replicate and leave the
fid.
On the PBworks production system, we ended up with 2.3TB of over-replicated
fids, that required the reconfiguration of the tracker to fix, and a multi-day
FSCK.
Original issue reported on code.google.com by martinb...@gmail.com on 9 Jul 2012 at 6:39
Original issue reported on code.google.com by
martinb...@gmail.com
on 9 Jul 2012 at 6:39