I just increased discovery.ec2.ping_timeout to 15s to avoid the split brain
situations I was getting ever so often, but according to [1] high
ping_timeout values increase the risk "that a new node’s discovery phase
will end before it has found the cluster". Isn't the purpose of increasing
the ping_timeout to avoid that?
Your original understanding is correct: discovery.ec2.ping_timeout
prevents and decreases the possibility that a node will leave the cluster
(due to noisy network, too many EC2 nodes, etc).
The sentence advices you to use tags (and/or security groups) for AWS
deployments with many, many other EC2 instances. Elasticsearch is able to
filter the nodes more effectively then.
Karel
On Saturday, January 5, 2013 3:56:13 AM UTC+1, Eric Jain wrote:
I just increased discovery.ec2.ping_timeout to 15s to avoid the split
brain situations I was getting ever so often, but according to [1] high
ping_timeout values increase the risk "that a new node’s discovery phase
will end before it has found the cluster". Isn't the purpose of increasing
the ping_timeout to avoid that?
The sentence advices you to use tags (and/or security groups) for AWS
deployments with many, many other EC2 instances. Elasticsearch is able to
filter the nodes more effectively then.
Thanks for the clarification. So shouldn't the docs read
"(particularly with low ping_timeout values)"?
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.