Number of Master eligible nodes

Elastic Stack Elasticsearch
1

I am trying to setup a large cluster 60nodes on 10physical servers (very
powerful servers). How many master eligible nodes should I setup? Would the
following be sufficient in this case?

10 master only nodes (no data) - One per physical machine
50 data only nodes (not master) - 5 per physical machine

discovery.zen.minimum_master_nodes: 6 (derived from ... 10/2 + 1 to solve
the split brain problem.)

Thanks.

--
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/470dea6c-0cf0-4346-8cc9-a991e49bf465%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

2

Without knowing the specs of the machines, that looks reasonable.

Regards,
Mark Walkom

Infrastructure Engineer
Campaign Monitor
email: markw@campaignmonitor.com
web: www.campaignmonitor.com

On 18 December 2013 03:52, Search User feedworks@gmail.com wrote:

I am trying to setup a large cluster 60nodes on 10physical servers (very
powerful servers). How many master eligible nodes should I setup? Would the
following be sufficient in this case?

10 master only nodes (no data) - One per physical machine
50 data only nodes (not master) - 5 per physical machine

discovery.zen.minimum_master_nodes: 6 (derived from ... 10/2 + 1 to solve
the split brain problem.)

Thanks.

--
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/470dea6c-0cf0-4346-8cc9-a991e49bf465%40googlegroups.com
.
For more options, visit https://groups.google.com/groups/opt_out.

--
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/CAEM624ZEn4VstaCHvoJiefvRR4mvW%3Db7X3tVifevq-Db5wbevg%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.

3

Start with 1 ES node per server. It makes not much sense to have 6 nodes
per server, even on "very powerful" servers.

Jörg

--
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/CAKdsXoE8hfev5mtsa0T4Trv3jXj0MqJQJQ24PyEyfOwKKKMYeQ%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.

4

Thanks Jorg. But can you please explain why it doesn't make sense? I
started with one and increased the data nodes. With more data nodes, I
started seeing better indexing throughput and my queries seem to be running
fast. I don't have numbers to give you but I can definitely run the
experiment and post.

Thanks

On Tuesday, December 17, 2013 8:31:47 PM UTC-5, Jörg Prante wrote:

Start with 1 ES node per server. It makes not much sense to have 6 nodes
per server, even on "very powerful" servers.

Jörg

--
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/ea7933b8-1d01-4e71-8a71-c916d27e2c61%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

5

If you install ES with default configuration, you install developer mode:
small heap, zen discovery, all reasonable feature enabled, many automatic
configurations for resource utilization.

You can change the ES configuration to adapt better to the RAM of your
server. ES scales to the number of CPU cores automatically. Also there are
recommendations for running JVM on large servers.

So, there is no need to run more than one ES JVM per server. Each JVM adds
overhead to the layer between ES und OS, and requires network communication
overhead - it must be slower than a single large node.

Jörg

--
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/CAKdsXoFHNgoZ%2Bc6-J51X5WeKusES%3DnNXN2aL2%3DXZ5UGxHu0a6g%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.