I am trying to upgrade ES cluster to 5 and using _plugin/elasticsearch-migration/ when i am running cluster checkup and its complaining about issue At least 65536 file descriptors must be available to Elasticsearch first before i can upgrade, Here is what my node stats looks
when i call _nodes/stats/process?filter_path=.max_file_descriptor i see below value
{"process":{"max_file_descriptors":-1}}}}
when i call _nodes/_local?process=true&pretty=true i see below value but not with process block
"max_file_descriptors" : "65536"
below entries in /usr/share/elasticsearch/elasticsearch.sh
Maybe just one node is reporting -1 as max file descriptors? Can you introspect the response of the node info api for all nodes? (Either by sending nodes/local?process=true to all nodes or nodes/stats/process to s single node (this should include the max file descriptions for all nodes in your cluster) ).
single node or cluster still it shows max_file_descriptor=-1,by the way it displays open files 65536 when i call ulimit -n, what options i have now to upgrade to 5 ?
Not sure why this happens. But you can easily verify if this is problematic for ES 5. ES 5 checks for this at startup and fails if it doesn't have enough file descriptors. So if you just start a temporary ES 5 node on your machine (don't bind to localhost, that will enable dev mode) and if it starts without a problem then I think the warning that migration plugin is showing you, can be ignored.
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.