I am running 1.0.2 with the default tiered compaction.  After running a "nodetool compact", I noticed that on about half of the machines in my cluster, both "nodetool ring" and "nodetool info" report that the load is actually higher than before when I expect it to be lower.  It is almost twice as much as before.  I did a du command on the data directory and found the the actual disk usage is only about half of what's being reported by nodetool.  Since I am running 1.0.2, there are no compacted sstables waiting to be removed.  I manually trigger a full GC in the JVM but that made no difference.  When I restarted Cassandra, nodetool once again report the correct load.

Is this a known problem?