Zookeeper has a strict quorum requirement that is hard to automate across 2 DC. 
You can make a manual procedure for how to make a secondary DC with minority of 
ZK nodes into the primary, with a quorum, by promoting observers to voting 
nodes. But fully automatic is worse.

If you can have one server in a 3rd DC/AZ just for the 3rd ZK node then you 
have a solution. Solr can still be in 2 DCs only.

Jan Høydahl

> 15. sep. 2021 kl. 13:28 skrev Christian Pfarr <[email protected]>:
> 
> Why should autoscaling not work for this scenario?
> 
> 
> Regards
> 
> Christian
> 
> 
> 
> 
> -------- Original-Nachricht --------
> Am 15. Sept. 2021, 13:25, Eric Pugh schrieb:
> 
> I don’t think that Solr really provides (yet!) a great solution across 
> datacenter.   I lean towards, if you want to run in multiple datacenters, 
> just run multiple fully independent clusters, and feed them from a common 
> queue.   That way, if something bad happens in one area, you don’t have the 
> other area trying to connect/communicate across a barrier.
> 
> 
> 
>> On Sep 15, 2021, at 6:02 AM, Christian Pfarr <[email protected]> wrote:
>> 
>> You can use autoscaling Feature to achieve this. We currently use it for 
>> balancing nodes within a Datacenter, but should also work with different AZ 
>> by providing a system property.
>> 
>> 
>> https://solr.apache.org/guide/8_7/solrcloud-autoscaling-policy-preferences.html#place-replicas-based-on-a-system-property
>> 
>> 
>> combine 3 AZ rules with 33% each .
>> 
>> 
>> https://solr.apache.org/guide/8_7/solrcloud-autoscaling-policy-preferences.html#multiple-percentage-rules
>> 
>> 
>> Not totally sure if it works that way, but you can give it a try.
>> 
>> 
>> Regards,
>> 
>> Christian
>> 
>> 
>> 
>> 
>> -------- Original-Nachricht --------
>> Am 15. Sept. 2021, 07:44, Walter Underwood schrieb:
>> 
>> You need three data centers. We split our Zookeeper ensemble across three 
>> AWS availability zones.
>> 
>> sunder
>> 
>> Sent from my iPad
>> 
>> > On Sep 14, 2021, at 10:28 PM, HariBabu kuruva <[email protected]> 
>> > wrote:
>> >
>> > Hi All,
>> >
>> > We have Solr cloud running on 10 nodes with Zookeeper running on 5 nodes.
>> > They are running across 2 data centers.
>> >
>> > As part of high availability we would like to run solr so that it should
>> > sustain even if one data center goes down.
>> >
>> > How can we achieve this approach as we need to maintain the Quorum rule for
>> > Zookeepers.
>> >
>> > Please advise
>> >
>> > --
>> >
>> > Thanks and Regards,
>> > Hari
>> > Mobile:9790756568
>> 
>> <publickey - EmailAddress([email protected]) - 0xF0E154C5.asc>
> 
> _______________________
> Eric Pugh | Founder & CEO | OpenSource Connections, LLC | 434.466.1467 | 
> http://www.opensourceconnections.com | My Free/Busy  
> Co-Author: Apache Solr Enterprise Search Server, 3rd Ed       
> This e-mail and all contents, including attachments, is considered to be 
> Company Confidential unless explicitly stated otherwise, regardless of 
> whether attachments are marked as such.
> 
> <publickey - EmailAddress([email protected]) - 0xF0E154C5.asc>

Reply via email to