Choking of Elasticsearch


(narinder.izap) #1

Hi there,

     We are using Elasticsearch 0.90.6 on our production server, with 

one replica. We need to import the data via scripts and we have million of
records to be saved. We are using the Bulk mode to save the documents to
Elasticsearch index.

Our architecture is using two Databases, One is Mysql, and other is
Elasticsearch. The data is first saved in Mysql, then it is sent to
Elasticseach for documents savings. But while data import script is
running, the data is being saved in Mysql, but not being pushed to
Elasticsearch, On investigating the nodes via Elastic HQ, I found the
following factors

https://lh4.googleusercontent.com/-smEWpFUSCV8/UzKvOwA2u6I/AAAAAAAAAF0/kxOtdJkcouw/s1600/esnodediagnostics.jpeg

     The filter eviction is in red colour, saying it has crossed the 

limit.I need to understand, the two terms used for this filter eviction(filter_cache.eviction
and query_total
), How are they being calculated and How can i control
them,How it is affecting the data saving and data searching too, or
anything else that can cause the document skipping from saving.The search
is also very very slow, although,we have given 30 GB of RAM to
Elasticsearch server.

Thanks in advance.
Narinder Kaur

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/e3aff1cf-d3ff-4037-ac3b-c846a0d00798%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


(Binh Ly-2) #2

This metric/warning should have no effect in indexing. The query_total is
the number of shard _searches executed so far, and filter cache evictions
counts the number of times something in the filter cache is
removed/replaced because you have reached the limits - this is affected by
filters in your queries. Filter cache evictions are normal as part of
running your queries, although the filter cache size can be increased (or
decreased) as needed. But before you do that, I'd probably find out first
why search is slow by analyzing your queries.

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/85f3e0aa-6b65-4376-9b33-a20ac287bfc6%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


(system) #3