Single NFS Storage for Entire Cluster - Separate processing and data replication

Elastic Stack Elasticsearch
1

I have a set of queries that take up a lot of RAM (mostly reads, few, infrequent writes), so I'm testing a cluster that has multiple nodes on different machines as query engines to field requests. However, my machines are already set with an NFS (without any local storage, only shared, and beyond my control). I realize that's kind of a weird topology abusing ES a bit. I'm trying to essentially separate the query processing engine (that I need multiple of) from the data distribution (that I can only have one of)---and am looking for settings that will help with this.
If my conf settings are all pointing to the same place:path.data = ~/data/elasticsearch
---my nodes are probably all overwriting on top of each other on start-up and all replication, right? One node undoing what the last node did? If this is the topology I'm stuck with, is there still a way to use ES? The settings I think I should have are:
cluster.routing.allocation.enable = noneindex.number_of_shards: 5index.number_of_replicas: 0
Is that correct? Are there other recommendations from you wizards? Thanks for your help!Bart

--
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/COL128-W11198BF1FEA92DAF1C5F3FA61B0%40phx.gbl.
For more options, visit https://groups.google.com/d/optout.

2

We don't recommend using NFS for storing ES data, but if that is all you
have then so be it.

They won't be overwriting, if you go to the data directory you will see it
creates a directory for each node and that node then stores data there.

You shouldn't need to do anything else.

On 9 March 2015 at 15:31, Bartleby O'Connor bartlebyoconnor@hotmail.com
wrote:

I have a set of queries that take up a lot of RAM (mostly reads, few,
infrequent writes), so I'm testing a cluster that has multiple nodes on
different machines as query engines to field requests. However, my machines
are already set with an NFS (without any local storage, only shared, and
beyond my control). I realize that's kind of a weird topology abusing ES
a bit. I'm trying to essentially separate the query processing engine
(that I need multiple of) from the data distribution (that I can only have
one of)---and am looking for settings that will help with this.

If my conf settings are all pointing to the same place:
path.data = ~/data/elasticsearch

---my nodes are probably all overwriting on top of each other on start-up
and all replication, right? One node undoing what the last node did? If
this is the topology I'm stuck with, is there still a way to use ES? The
settings I think I should have are:

cluster.routing.allocation.enable = none
index.number_of_shards: 5
index.number_of_replicas: 0

Is that correct? Are there other recommendations from you wizards?
Thanks for your help!
Bart

--
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/COL128-W11198BF1FEA92DAF1C5F3FA61B0%40phx.gbl
https://groups.google.com/d/msgid/elasticsearch/COL128-W11198BF1FEA92DAF1C5F3FA61B0%40phx.gbl?utm_medium=email&utm_source=footer
.
For more options, visit https://groups.google.com/d/optout.

--
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/CAEYi1X_aZVuKk38p_ZTr3C%3D%2B0hH9hZv35%3DArmkkT69gudgUhuw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.