gearpump-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (GEARPUMP-83) After killing all worker instances, application status should not be described as active
Date Thu, 02 Jun 2016 07:31:59 GMT

    [ https://issues.apache.org/jira/browse/GEARPUMP-83?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15311879#comment-15311879
] 

ASF GitHub Bot commented on GEARPUMP-83:
----------------------------------------

Github user huafengw commented on a diff in the pull request:

    https://github.com/apache/incubator-gearpump/pull/31#discussion_r65495393
  
    --- Diff: daemon/src/main/scala/org/apache/gearpump/cluster/master/AppManager.scala ---
    @@ -310,22 +320,20 @@ private[cluster] class AppManager(kvService: ActorRef, launcher:
AppMasterLaunch
       case class RecoverApplication(applicationStatus: ApplicationState)
     
       private def cleanApplicationData(appId: Int): Unit = {
    -    // Add the dead app to dead appMaster
    -    appMasterRegistry.get(appId).foreach { pair =>
    -      val (appMasterActor, info) = pair
    -      deadAppMasters += appId -> (appMasterActor, info.copy(
    -        finishTime = System.currentTimeMillis()))
    -    }
    -
    -    appMasterRegistry -= appId
    +    // Add the dead app to dead appMasters
    +    deadAppMasters += appId
    --- End diff --
    
    When the application is down, you don't update its finish time. You can see the dashboard
missed that info.


> After killing all worker instances, application status should not be described as active
> ----------------------------------------------------------------------------------------
>
>                 Key: GEARPUMP-83
>                 URL: https://issues.apache.org/jira/browse/GEARPUMP-83
>             Project: Apache Gearpump
>          Issue Type: Bug
>          Components: Dashboard
>    Affects Versions: 0.8.0
>            Reporter: Kam Kasravi
>            Assignee: Manu Zhang
>            Priority: Minor
>             Fix For: 0.8.1
>
>
> Step to reproduce:
> Start cluster with one worker
> Start a word count
> Kill the worker
> Expect /api/v1.0/master/applist actually returns app status as active, but application's
detail page is not available. I think as there is no resource to run the application, the
application is in some abnormal status. In order not to mislead user, I think we should invent
a new status, might be recovering or something.
> Example output:
> {code}
> {"appMasters":[{"status":"active","appId":1,"appName":"dag","appMasterPath":"akka.tcp://app1-executor-1@127.0.0.1:46761/user/daemon/appdaemon1/$c","workerPath":"akka.tcp://48a47aa6-81c0-493c-9948-9d7d4c946db6@127.0.0.1:59201/user/Worker48a47aa6-81c0-493c-9948-9d7d4c946db6","submissionTime":"1451894551477","startTime":"1451894553568","user":"qxu"},{"status":"active","appId":2,"appName":"wordCount","appMasterPath":"akka.tcp://app2-executor-1@127.0.0.1:49261/user/daemon/appdaemon2/$c","workerPath":"akka.tcp://48a47aa6-81c0-493c-9948-9d7d4c946db6@127.0.0.1:59201/user/Worker48a47aa6-81c0-493c-9948-9d7d4c946db6","submissionTime":"1451898038991","startTime":"1451898040265","user":"qxu"}]}
> {code}



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

Mime
View raw message