[ 
https://issues.apache.org/jira/browse/SLING-10709?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17396244#comment-17396244
 ] 

Timothee Maret commented on SLING-10709:
----------------------------------------

Thanks [~cziegeler] for the PR.

The purpose of the `SubscriberIdleCheck` is to expose a publish service to 
traffic (via the LB) only when it has catch up processing the latest messages 
on the journal. The check is based on an heuristic. It is a best effort 
mechanism that turns green in any case after [5 
minutes|https://github.com/apache/sling-org-apache-sling-distribution-journal/blob/697c9dd7ba83c5bf10e6a148aba971db1e8e7832/src/main/java/org/apache/sling/distribution/journal/impl/subscriber/SubscriberIdle.java#L34].
{quote}Previous status RED is mapped to CRITICIAL (not sure if this is the 
right thing)
{quote}
Looking at the HC 
[readme|https://github.com/apache/felix-dev/blob/master/healthcheck/README.md], 
it seems that CRITICAL status can trigger re-provisioning the system from 
scratch. I think that the measure is no adapted and can lead to never ending 
re-provisioning cycles in case new messages keep coming on the journal.

The semantic of the TEMPORARILY_UNAVAILABLE status seems to fit the purpose of 
the SubscriberIdleCheck better. We'd need to configure the 
`temporarilyAvailableGracePeriodInMs` configuration to something larger than 5 
minutes which seems to be the case by default. 

 

> Migrate SubscriberIdleCheck  from Systemready to HC
> ---------------------------------------------------
>
>                 Key: SLING-10709
>                 URL: https://issues.apache.org/jira/browse/SLING-10709
>             Project: Sling
>          Issue Type: Improvement
>          Components: Content Distribution
>    Affects Versions: Content Distribution Journal Core 0.1.18
>            Reporter: Carsten Ziegeler
>            Priority: Major
>             Fix For: Content Distribution Journal Core 0.1.20
>
>
> Apache Felix systemready is deprecated in favour of Apache Felix HealthCheck. 
> This component should be migrated



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to