ambari-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hudson (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (AMBARI-9128) Bulk Ops on slaves for a service does not work as expected when the service is in maintenance mode
Date Wed, 14 Jan 2015 19:08:36 GMT

    [ https://issues.apache.org/jira/browse/AMBARI-9128?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14277472#comment-14277472
] 

Hudson commented on AMBARI-9128:
--------------------------------

FAILURE: Integrated in Ambari-trunk-Commit-docker #757 (See [https://builds.apache.org/job/Ambari-trunk-Commit-docker/757/])
AMBARI-9128. Bulk Ops on slaves for a service does not work as expected when the service is
in maintenance mode (alexantonenko) (hiveww: http://git-wip-us.apache.org/repos/asf?p=ambari.git&a=commit&h=32f1b05fdf12150a23ad05616377fa41d25fd8aa)
* ambari-web/app/controllers/main/host.js


> Bulk Ops on slaves for a service does not work as expected when the service is in maintenance
mode
> --------------------------------------------------------------------------------------------------
>
>                 Key: AMBARI-9128
>                 URL: https://issues.apache.org/jira/browse/AMBARI-9128
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-web
>    Affects Versions: 2.0.0
>            Reporter: Antonenko Alexander
>            Assignee: Antonenko Alexander
>             Fix For: 2.0.0
>
>         Attachments: AMBARI-9128.patch
>
>
> ###
> I am observing inconsistency in Ambari’s behavior (1.7.0) and would like to see if
we should file a bug. My Hbase is in the maintenance mode with 2 RS down.
> Next, I go to the service -> click RS link -> explicitly select all hosts with
RS servers in the Hosts screen -> Actions -> Start RS.
> Ambari even pops up a confirmation screen, but then doesn’t initiate any action.
> I understand that maintenance mode prevents components from responding to bulk requests,
but the behavior I described is wrong – we can’t expect an admin go an start a component
on each host manually.
> I’d expect this – components won’t respond to bulk requests, but if I explicitly
select hosts and perform an action – it should go through. Especially if Ambari asks me
for a confirmation.
> ###
> This should go thru.  HBase is in Maintenance Mode, but specific operations on HBase
itself or its slaves should work.
> PUT request to '/api/v1/clusters/HDP_2_2/host_components' sets an operation level in
RequestInfo to CLUSTER. 
> {code}
> {
> 	"RequestInfo":{
> 		"context":"Start RegionServers",
> 		"operation_level":{"level":"CLUSTER","cluster_name":"HDP_2_2"},
> 		"query":"HostRoles/component_name=HBASE_REGIONSERVER&HostRoles/host_name.in(c6403.ambari.apache.org,c6407.ambari.apache.org)&HostRoles/maintenance_state=OFF"
> 	},
> 	"Body":{
> 		"HostRoles":{"state":"STARTED"}
> 	}
> }
> {code}
> But our host-component maintenance state is IMPLIED_FROM_SERVICE.
> So our API rejects state change.
> UI should set an operation_level level to SERVICE in this case.
> {{(Actions -> Selected/Filtered/All hosts -> RegionServers/DataNodes... etc)}}
> So new PUT request should looks like:
> {code}
> {
> 	"RequestInfo":{
> 		"context":"Start RegionServers",
> 		"operation_level":{"level":"SERVICE","cluster_name":"HDP_2_2"},
> 	.......
> }
> {code}



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

Mime
View raw message