Hello, I have a situation when I need to shutdown a node, and then
start it up after 5 minutes. The node is a datanode, what I see is
that after the startup the datanode will basically wait for the
cluster to become green without recovering any of its local shards
first. This causes excessive replication, then rebalancing when
cluster is green back to the node that has started, so we end up
rewriting data (slushing) several times. How can I tune my cluster to
have a restarted datanode start local recovery right away?
On Thursday, March 29, 2012 3:10:54 AM UTC+8, Jack Levin wrote:
Hello, I have a situation when I need to shutdown a node, and then
start it up after 5 minutes. The node is a datanode, what I see is
that after the startup the datanode will basically wait for the
cluster to become green without recovering any of its local shards
first. This causes excessive replication, then rebalancing when
cluster is green back to the node that has started, so we end up
rewriting data (slushing) several times. How can I tune my cluster to
have a restarted datanode start local recovery right away?
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.