Hi Dong, I¹m asking Tao whether the ³QUERY² mode instances can get the ³Cube² status change automatically (not job status);
For example, I create a new cube, its initial status is ³Disabled²; Then I trigger a build job, which will be executed in the ³job engine² instance; When the job build is completed, the job engine will update this cube¹s status to ³Active², and also send REST calls to the instances in ³kylin.rest.servers²; The instances in ³kylin.rest.servers² will flush their caches on receiving this REST call, so they will get the cube¹s latest status ³Active²; All these things happen automatically; If you observed the the ³query² mode nodes didn¹t get the cube¹s state change, while the ³job engine² instance got successfully, that indicates the configuration may be wrong for other instances; because the ³job engine² node also depends this REST call to flush the cache in itself; If all your instances are in the same machine, you can use localhost as the hostname, I¹m not sure this can solve your problem, but you can give a try; If the problem is still there, welcome to open a JIRA for us. Thanks for the input to Kylin. On 5/6/15, 4:23 PM, "dong wang" <[email protected]> wrote: >yes, all the "QUERY" mode instances got the job status for the cube >correctly, and as checked the >kylin.rest.servers=all-machine:por1,query-1-machine:port2,query-3-machine: >port3 >should be correct as well
