[ 
https://issues.apache.org/jira/browse/SLING-5058?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Stefan Egli updated SLING-5058:
-------------------------------
    Fix Version/s:     (was: Discovery Impl 1.2.0)
                   Discovery Impl 1.2.2

> introduce viewCnt to ./establishedView to be able to detect missing changes
> ---------------------------------------------------------------------------
>
>                 Key: SLING-5058
>                 URL: https://issues.apache.org/jira/browse/SLING-5058
>             Project: Sling
>          Issue Type: Improvement
>          Components: Extensions
>    Affects Versions: Discovery Impl 1.1.6
>            Reporter: Stefan Egli
>            Assignee: Stefan Egli
>            Priority: Minor
>             Fix For: Discovery Impl 1.2.2
>
>
> With SLING-5030 when an instance detects that it is ISOLATED_FROM_TOPOLOGY it 
> will resets its {{leaderElectionId}} to ensure that it is not becoming leader 
> again. This works fine when the instance actually detects that it is 
> isolated. If the timing however is such that it does not detect the isolation 
> (eg when the isolation is for a very short time only), then it would not be 
> able to reset the {{leaderElectionId}}, thus it might become leader again - 
> which means that we have an unnecessary leader switch.
> This could be improved by introducing a {{viewCnt}} on the 
> {{./establishedView}} - and an instance could detect a gap in the viewCnt and 
> *then* reset the {{leaderElectionId}}.
> Putting this out-of-scope for discovery.impl 1.1.8 though



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

Reply via email to