When transferring shards, the network speed between 2 nodes can increase up to 16 G/s. I want to limit network usage to 7G/s for the entire cluster.
Can this be done by means of ElasticSearch?
The only network regulator I could find is indices.recovery.max_bytes_per_sec, but it doesn't help
The default maximum speed is made up of indices.recovery.max_bytes_per_sec which defaults at 40mb/s per index, and this is multiplied by cluster.routing.allocation.node_concurrent_recoveries which default to 2 (incoming or outgoing), so once cluster is green and rebalance is needed/started, the default speed for shard rebalancing should be around 80Mb/s (2 outgoing or incoming * 40mb/s)
I don't think that's right, indices.recovery.max_bytes_per_sec applies in aggregate across all recoveries. Adding concurrent recoveries will often lengthen the overall recovery time.
Thank you so 40mb is limit per node shared by all recovery per node, concurrency is detrimental especially on spinning disks so best to increase speed than concurrency
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.