Re: Is it possible to configure a failover backend?

[Marco Corte]

Hi

I would try Jarno's method with the "frontend" rules: simple, effective, 
great.


Similar results may achieved with the "backup"s inside each "backend" 
section.
Because of the "use_backend" map and a ton of other reasons, this 
approach should not fit your case, but... who knows?


  backend loggingdc1-elasticsearch-acc
  mode http
  balance roundrobin
  option httplog
  option httpchk GET /
  option allbackups
  server loggingdc1-elasticsearch-01a:9200 10.1.20.22:9200 check
  server loggingdc1-elasticsearch-03a:9200 10.1.20.23:9200 check
  server loggingdc1-elasticsearch-05a:9200 10.1.20.25:9200 check
  server loggingdc2-elasticsearch-02a:9200 10.1.36.23:9200 check backup
  server loggingdc2-elasticsearch-04a:9200 10.1.36.25:9200 check backup
  server loggingdc2-elasticsearch-06a:9200 10.1.36.26:9200 check backup

Ciao!

.marcoc

Re: Is it possible to configure a failover backend?

[Jarno Huuskonen]

Hi,

On Mon, Jan 29, Pieter Vogelaar wrote:
> I have two Elasticsearch clusters. One is the active cluster and the other is 
> the failover cluster.
> 
> Is it possible to configure a failover backend if the default backend health 
> check failes?

Does "if the default backend health check failes" mean that all servers
in default backend are down ?

nbsrv + acl + use_backend
(https://cbonte.github.io/haproxy-dconv/1.8/configuration.html#7.3.1-nbsrv)

frontend
 acl primary_down nbsrv(loggingdc1-elasticsearch-acc) lt 1
 use_backend loggingdc2-elasticsearch-acc if primary_down
 use_backend loggingdc1-elasticsearch-acc if !primary_down
 (or default_backend loggingdc1-elasticsearch-acc)

-Jarno

> So if loggingdc1-elasticsearch-acc is not healthy, I would like to route to 
> the loggingdc2-elasticsearch-acc backend.
> 
> 
> The configuration:
> 
> frontend default-http
>   bind :80
>   bind :8091
>   bind :9200
>   mode http
>   option httplog
>   use_backend 
> %[base,regsub(^www\.,,i),map_beg(/etc/haproxy/http-bases-to-backends.map,default)]
> 
> backend loggingdc1-elasticsearch-acc
>   mode http
>   balance roundrobin
>   option httplog
>   option httpchk GET /
>   server loggingdc1-elasticsearch-01a:9200 10.1.20.22:9200 check
>   server loggingdc1-elasticsearch-03a:9200 10.1.20.23:9200 check
>   server loggingdc1-elasticsearch-05a:9200 10.1.20.25:9200 check
> 
> backend loggingdc2-elasticsearch-acc
>   mode http
>   balance roundrobin
>   option httplog
>   option httpchk GET /
>   server loggingdc2-elasticsearch-02a:9200 10.1.36.23:9200 check
>   server loggingdc2-elasticsearch-04a:9200 10.1.36.25:9200 check
>   server loggingdc2-elasticsearch-06a:9200 10.1.36.26:9200 check
> 
> 
> Best regards,
> Pieter Vogelaar

-- 
Jarno Huuskonen

Re: Is it possible to configure a failover backend?

[Paul Lockaby]

Yes. Whichever backend server you do not want to route requests to except when 
the primary server is offline, append the keyword "backup" to that server's 
configuration (after the word "check").

E.g.:

server loggingdc1-elasticsearch-05a:9200 10.1.20.25:9200 check backup

-Paul

> On Jan 29, 2018, at 4:58 AM, Pieter Vogelaar  wrote:
> 
> Hi,
>  
> I have two Elasticsearch clusters. One is the active cluster and the other is 
> the failover cluster.
>  
> Is it possible to configure a failover backend if the default backend health 
> check failes?
> So if loggingdc1-elasticsearch-acc is not healthy, I would like to route to 
> the loggingdc2-elasticsearch-acc backend.
>  
>  
> The configuration:
>  
> frontend default-http
>   bind :80
>   bind :8091
>   bind :9200
>   mode http
>   option httplog
>   use_backend 
> %[base,regsub(^www\.,,i),map_beg(/etc/haproxy/http-bases-to-backends.map,default)]
>  
> backend loggingdc1-elasticsearch-acc
>   mode http
>   balance roundrobin
>   option httplog
>   option httpchk GET /
>   server loggingdc1-elasticsearch-01a:9200 10.1.20.22:9200 check
>   server loggingdc1-elasticsearch-03a:9200 10.1.20.23:9200 check
>   server loggingdc1-elasticsearch-05a:9200 10.1.20.25:9200 check
>  
> backend loggingdc2-elasticsearch-acc
>   mode http
>   balance roundrobin
>   option httplog
>   option httpchk GET /
>   server loggingdc2-elasticsearch-02a:9200 10.1.36.23:9200 check
>   server loggingdc2-elasticsearch-04a:9200 10.1.36.25:9200 check
>   server loggingdc2-elasticsearch-06a:9200 10.1.36.26:9200 check
>  
>  
> Best regards,
> Pieter Vogelaar