I have three node cluster(1 master node and two data node). currently i am creating weekly index with 3 shards and 1 replica.and size of weekly index is around 40-50 gb. and its working fine for us.
but now we have another new requirement with daily log size of around100gb (it can grow upto 500gb after 2-3 months) . i have tried with weekly index but its creating performance issue(slowness, timeout, etc)
Can you please suggest me that should i increase number of shards or create daily basis index with 3 shards.
This is a bad setup. You should always have at least 3 master-eligible nodes, so I would recommend changing this to 3 master/data nodes or 1 dedicated master node and 2 master/data nodes).
The easy way to manage changing volumes, especially for time-series data that is not updated, is to use rollover and ILM. Rollover allows you to specify maximum size and age and then create a new index whichever comes first. When you have volumes that do not trigger the size threshold you can get weekly indices and once volumes ramp up each index will cover a shorter time period and instead aim for the target size.
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.