FYI,
I blogged about how to use a mod_proxy_balancer to use multiple
Karaf/ServiceMix instances and "detect" the ones up:
http://blog.nanthrax.net/2013/02/load-balancing-with-apache-karaf-cellar-and-mod_proxy_balancer/
In this blog, I use the mod_proxy_balancer module from Apache Httpd, but
I did similar config in LB appliance (F5, Juniper, Cisco).
Regards
JB
On 05/27/2014 06:57 PM, Dominique Broeglin wrote:
Hello!
On a setup I’m currently working on we use « ping » URLs for the load balancer.
This way the loadbalancer knows that the service is available if it gets a 200
OK on /svc/service/ping. If the service is not running it would get either a
connection refused, at 404 or even a 500.
Best regards,
Dominique
On 23 May 2014, at 17:08 , Christian Müller <[email protected]> wrote:
Hello guys!
We are using Apache ServiceMix (Fuse ESB 7.1.0) in a master/slave
configuration. We are providing some web services in our application (based
on Apache Camel) by using the OSGI servlet service (central jetty.xml
configuration). Everithing works fine so far.If the instance 1 (master)
goes down, the instance 2 (slave) will take over. Our load balancer will
detect this and forward the requests to instance 2 (now master).
Our issue is, if we start instance 1 again (now slave), the load balancer
detects the http port is available again (but not the service behind) and
forward the requests to instance 1 (which cannot handle the requests
because he is the slave).
Any recommendations how we should handle this?
We do not use any web console in ServiceMix.
Thanks in advance,
Christian
--
Jean-Baptiste Onofré
[email protected]
http://blog.nanthrax.net
Talend - http://www.talend.com
