This means that the search queue on your remote (where the follower shard is pulling from) is at capacity, and rejected the search requests the follower runs on the remote cluster to find the changes that it needs to replicate. This means that your remote cluster is overloaded.
Thanks @jasontedor. I wanted to understand one more thing. Currently we have an ELB in front of Elastic Data nodes (ELB:9200). For replication, we need to apply cluster settings something like below:
curl -X PUT "10.167.65.123:9200/_cluster/settings?pretty" -H 'Content-Type: application/json' -d'
{
"persistent" : {
"cluster" : {
"remote" : {
"us_east_cluster" : {
"seeds" : [
"10.24.21.168:9300"
]
}
}
}
}
}
'
So can we use ELB:9300 for seeds instead of node IP's since they are not reliable? Currently I am not sure, if the replication will break if the node IP's get refreshed within the cluster.
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.