Es 5.6.x force merege question about max_num_segments/max_merge_count/max_thread_count

Elastic Stack Elasticsearch
1

i see the doc have setting with the indices about force merege
index.merge.scheduler.max_merge_count
index.merge.scheduler.max_thread_count
with not ssd the docs give index.merge.scheduler.max_thread_count to "1" ,and the
index.merge.scheduler.max_merge_count aslo need set to 1

the question is :

  • 1.the max_num_segments in the flow cmd when force merege is same as max_merge_count?
    curl -XPOST "http://xxxxx:9200/xxxx/_forcemerge?max_num_segments=1
  • 2.have many segments better with index search or small segment better with index search?
2

with monitoring
the one index have so many segment count
image

5

some one can help me?

6

i test it. can't find how setting about it........

7

there no doc about this..
i see the soure code about this .


only with +5 ...
i don't know which is better about index.merge.scheduler.max_merge_count and index.merge.scheduler.max_thread_count and max_num_segments for kibana search.
the index have 4.3b docs..
image
image.png1512×682 63.5 KB

8

I'm not sure what your question is, but you'll find the documentation about _forcemerge and its request parameters here.

In general Elasticsearch will handle its own segment merging well, but it's sometimes useful to trigger a forcemerge as it will expunge deleted documents when merging two segments into one and therefore save you disk space.

But note that _forcemerge should not be called on an active index as that may result in huge (>5G) segments that will usually never be merged again and therefore could build up many deleted documents if that shard is still being updated.

9

about the this i want to know about which number is better as flow
"index.merge.scheduler.max_merge_count": "20",
"index.merge.scheduler.max_thread_count": "2"
i want to know ,use kibana search the index before _forcemerge?max_num_segments=1 is good or after forcemerge?

and the max_num_segments can set as "2" or other greater than 1 number ,which is better for search index data?

10

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.