Aleksandr Kovalenko created AMBARI-16746:
--------------------------------------------

             Summary: Changing and saving the llap percentage slider does not 
change the YARN queue configuration       
                 Key: AMBARI-16746
                 URL: https://issues.apache.org/jira/browse/AMBARI-16746
             Project: Ambari
          Issue Type: Bug
            Reporter: Aleksandr Kovalenko
            Assignee: Aleksandr Kovalenko
            Priority: Critical


As solution, lets do yarn queue refresh when user explicitly triggers "Restart 
LLAP" custom action from service actions menu or under Hive Server Interactive 
host component action menu on host details page. This will make sure that when 
restart llap happens, it always has the latest capacity-scheduler configs.

We can also optimize to do additional yarn queue refresh only when yarn configs 
are found to be stale. At this point lets use request_schedule API to batch 
"Refresh yarn queue" and "Restart LLAP" in single API call, so this can sustain 
browser refreshes and does not require the additional ui logic for persisting 
request state in the local DB or persist API endpoint. We can use following API:

{code}
curl 'http://c6401.ambari.apache.org:8080/api/v1/clusters/c1/request_schedules' 
--user admin:admin -H "X-Requested-By: ambari" -i -X POST  --data 
'[{"RequestSchedule":{"batch":[{"requests":[{"order_id":1,"type":"POST","uri":"/api/v1/clusters/c1/requests","RequestBodyInfo":{"RequestInfo":{"context":"Refresh
 YARN Capacity 
Scheduler","command":"REFRESHQUEUES","parameters/forceRefreshConfigTags":"capacity-scheduler"},"Requests/resource_filters":[{"service_name":"YARN","component_name":"RESOURCEMANAGER","hosts":"c6401.ambari.apache.org"}]}},{"order_id":2,"type":"POST","uri":"/api/v1/clusters/c1/requests","RequestBodyInfo":{"RequestInfo":{"context":"Restart
 
LLAP","command":"RESTART_LLAP"},"Requests/resource_filters":[{"service_name":"HIVE","component_name":"HIVE_SERVER_INTERACTIVE","hosts":"c6401.ambari.apache.org"}]}}]},{"batch_settings":{"batch_separation_in_seconds":1,"task_failure_tolerance":0}}]}}]'
{code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to