I am not aware of any issues with the max_snapshot_bytes_per_sec setting.
To check/verify whether the setting being ignored is truly the issue here, you could set it to a lower value and check if the snapshots become slower.
More likely the bottle neck is in the write throughput you can achieve to your NFS system or the rate at which your data nodes are able to read the data at.
If you set the max snapshot write rate to 1G this means that the disks the data is read from have to provide 1G, your network has to be able to handle 1G and the NFS must be able to sustain writes at that rate.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.