I have two kind of jobs. Jobs that runs periodically every night and jobs that come on demand from the client. I implemented the HA for oozie so I have my VIP.
I have two machines running one oozie server in each
My question is:
Does it have sense if I configure OOZIE_URL for the scheduled jobs for each master as localhost:11000/oozie and use the VIP load-balancer:11000/oozie only for the jobs that come from the client on demand ?
I would suggest not to complicate things. If your load balancer configuration is working fine, I would suggest to use OOZIE_URL=http://<load-balancer>:11000/oozie.
Anyways your load balancer is going to forward your request to one of the Oozie server so I would suggest to go with load balancer URI.
But it isn't weird that my oozie server number 1 use the load balancer to run a job, and then the load balancer gives back the ip of the oozie server number 1 ( assuming the the oozie server number 2 is down or saturated)?
Is it the load balancer or VIP made to give a single point of access to the client so he doesn't need to check which server is alive ?
I would like to know what are the disadvantage of the conf, like for example, I have the risk to have 2 jobs scheduled running on same time?
hi @Kuldeep Kulkarni, just a follow up question here. I just wondering what should I use on the oozie_url=http://<load-balancer>:11000/oozie. Is this some hostname I need to define in my DNS config and to which ip should i point it? Or i just need to leave it as default (pointing to hiveserver2 number 1). Thanks in advanced!