[jira] [Commented] (SENTRY-1752) HMSFollower gets stuck once it fails to process a notification event

["Hadoop QA (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SENTRY-1752?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16035611#comment-16035611 ] 

Hadoop QA commented on SENTRY-1752:
-----------------------------------

Here are the results of testing the latest attachment
https://issues.apache.org/jira/secure/attachment/12871061/SENTRY-1752.005-sentry-ha-redesign.patch against sentry-ha-redesign.

{color:green}Overall:{color} +1 all checks pass

{color:green}SUCCESS:{color} all tests passed

Console output: https://builds.apache.org/job/PreCommit-SENTRY-Build/2801/console

This message is automatically generated.

> HMSFollower gets stuck once it fails to process a notification event
> --------------------------------------------------------------------
>
>                 Key: SENTRY-1752
>                 URL: https://issues.apache.org/jira/browse/SENTRY-1752
>             Project: Sentry
>          Issue Type: Sub-task
>          Components: Sentry
>    Affects Versions: sentry-ha-redesign
>            Reporter: Na Li
>            Assignee: Na Li
>            Priority: Critical
>              Labels: security
>             Fix For: sentry-ha-redesign
>
>         Attachments: SENTRY-1752.001-sentry-ha-redesign.patch, SENTRY-1752.002-sentry-ha-redesign.patch, SENTRY-1752.003-sentry-ha-redesign.patch, SENTRY-1752.004-sentry-ha-redesign.patch, SENTRY-1752.004-sentry-ha-redesign.patch, SENTRY-1752.004-sentry-ha-redesign.patch, SENTRY-1752.005-sentry-ha-redesign.patch
>
>
> If HMSFollower encounters an event that causes exception, it just stops processing for this run and does not continue processing the rest of notification events. 
> The end result is that it cannot process any more notification events, and Sentry won't get any update from meta store from then on.
> for example, due to bug SENTRY-1750 "HMSFollower does not handle view update correctly", A create view event will cause exception. And the processing ends for this run. Next run, this event is still fetched from meta store, exception happens, and it stops processing again. So no events from that failed event can be processed, and no more update will be picked up be Sentry.
> When there is an exception in processing a notification event, should we still update the current notification ID, so we can move on?



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