lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sameer Maggon <>
Subject Re: New to Solr - Need advice on clustering
Date Tue, 26 Nov 2013 05:57:39 GMT

Take a look at Solr Replication. Essentially, you'll treat one as a master
& one as a slave. Both master & slave can be used to serve traffic. If one
of them goes down, the other can be used as a master for the interim.


On Mon, Nov 25, 2013 at 9:50 PM, Anders Kåre Olsen <> wrote:

> Hi Gora
> Thank you for your reply.
> We are planning on having a loadbalancer in front of our frontend servers.
> If I have two distinct solr indexes, how will I keep them synchronized? I
> expect that one of the frontend servers will have the task of updating the
> product repository on the e-commerce site. This server will then update the
> local solr index after product update has finished.
> Is there an easy  way that I can keep the two indexes synchronized without
> solrcloud?
> Regards
> Anders
> -----Oprindelig meddelelse----- From: Gora Mohanty
> Sent: Tuesday, November 26, 2013 2:37 AM
> To:
> Subject: Re: New to Solr - Need advice on clustering
> On 26 November 2013 01:44, Anders Kåre Olsen <> wrote:
>> Hi Solr-users
>> I’m trying to setup Solr for search and indexing on the project I’m
>> working on.
>> My project is a e-commerce B2B solution. We are planning on setting up 2
>> frontend servers for the website, and I was planning on installing Solr on
>> these servers. We are using Windows Server 2012 for the frontend servers.
>> We are not expecting a huge load on the servers, so we expect these 2
>> servers to be adequate to handle both the website and search index.
>> I have been looking at SolrCloud and ZooKeeper. Howver I have read that
>> you need at least 3 ZooKeepers in an ensamble, and I only have 2 servers.
>> I need to handle the situation where one of the servers crashes, so I
>> need both servers to have a Solr index.
> [...]
> If you do not want to get into SolrCloud, a simpler
> solution might be a HTTP load balancer in front of
> the two Solr instances. Hardware load balancers are
> better, but more expensive. A software load balancer
> like haproxy should meet your needs.
> Regards,
> Gora

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message