i have a three node es7.4 cluster without data. choose one node,ifdown the network,and es filehandler increasing quickly until the ulimit filehandler fills up。the new added filehandler point the same socket id as:ll /proc/esid/fd
965 -> socket:[12437267]
967 -> socket:[12437267]
...
all new filehanler point to socket:[12437267]
and socket:[12437267] is same to command "netstat -anp|grep espid" output:
unix 2 STREAM COMMECTED 12437267 espid/java
yes,as network is recoveryed, the es service can not be auto recoveryed because the file handler has filled up.es7.4.0 indeed has this problem,but es7.10.0 is ok。
also wish reply,many thanks
Both 7.4 and 7.10 are very old versions, long past EOL, and missing out on literally years of new features and bug fixes. You need to upgrade to a supported version as soon as possible.
thanks for your reply! I will upgrade version as soon as possible. but want to know the general reasons leading to too many file handler using the same socket with type of unix.
is it possible that a leak of channel connecting itself or other nodes?
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.