I am looking at creating an ES 3 node cluster all master and data in Windows on 3 hyper-V VM's. I have previously built a test environment in linux and attached 3 x separate 1TB storage drives for each node to replicate the cluster data.
I am wondering do I need to do the same for windows or could I attach a shared storage cluster to the VM so that the 3 nodes would then use and see the same data?
So it would be best to do exactly the same as my test environment and attach an extra 1TB storage drive to save and replicate the data for each node?
Out of interest, is there any 'gotcha's' I need to consider when building on Windows, I am more familiar supporting a windows environment, but not opposed to using Linux in production if it works much better with ElasticSearch?
This is not what elasticsearch is intended for and I think its not possible.
Yes.
Why do you have a problem with this approach? The good thing with elasticsearch is the data redundancy, if one of your harddrives dies you will have 2 other drives that still hold all the indices.
Elasticsearch works fine on windows and I had no problems with it. The only tricky thing that I couldnt get to work was starting it as a service. I tried all tutorials that I could fine but the cluster wouldn't connect. Now I use The NSSM (Non sucking serice manager) and it works perfectly.
I asked the same question some time ago. They both work very well:
The virtual servers will be built on a SAN with it's own clustering diskspace. I was just wondering if that was an option to use, rather than carving up the diskspace for each node. I have no problem with this option, I just need to pass this information onto my manager and he asked the question whether that was an option. I just wanted to rule it out 100%, so that's great.
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.