[jira] [Updated] (AMBARI-20593) EU/RU Auto-Retry does not reschedule task when host is not heartbeating before task is scheduled and doesn't have a start time

["Alejandro Fernandez (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/AMBARI-20593?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Alejandro Fernandez updated AMBARI-20593:
-----------------------------------------
    Summary: EU/RU Auto-Retry does not reschedule task when host is not heartbeating before task is scheduled and doesn't have a start time  (was: RU Auto-retry does not start when Restarting NN Batch 2 step is corrupted [Batch 1 was corrupted and fixed before])

> EU/RU Auto-Retry does not reschedule task when host is not heartbeating before task is scheduled and doesn't have a start time
> ------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: AMBARI-20593
>                 URL: https://issues.apache.org/jira/browse/AMBARI-20593
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-server
>    Affects Versions: 2.5.0
>         Environment: rolling upgrade
>            Reporter: Sviatoslav Tereshchenko
>              Labels: rolling_upgrade
>             Fix For: 2.5.1
>
>
> STR:
> 1) Install ambari 2.5.0.1
> In the ambari.properties file, set
> stack.upgrade.auto.retry.timeout.mins=6
> stack.upgrade.auto.retry.check.interval.secs=30
> 2) Install HDP with any set of services
> 3) Add NameNode HA
> 4) Register and install new HDP stack version
> 5) Start RU
> 5) Corrupt one step from Core Masters group (e.g., stop ambari-agent on a node while the command is running)
> Ambari will restart Restarting NN Batch 1 
> 6) Fix corrupted step (e.g., start ambari-agent again)
> 7) Corrupt another step from before the command is scheduled (e.g., stop ambari-agent on a node)
> 8) Fix corrupted step (e.g., start ambari-agent agent)
> The expectation is that Ambari Server should schedule the command on the 2nd node. However, because the command never got an original_start_time and start_time, the RetryUpgradeActionService was not able to retry it since it didn't have any timestamps to compare against.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)