[
https://issues.apache.org/jira/browse/OAK-1084?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13804671#comment-13804671
]
Jukka Zitting commented on OAK-1084:
------------------------------------
With the OAK-1113 change in place, each listener has its own event thread. If
that thread is unable to keep up with incoming changes, Oak will automatically
stop delivering changes to that listener. Once the event queue of that listener
is no longer full, any pending changes will get reported as a single big commit
with no local commit information (user data, etc.).
> Allow blacklisting of rogue observation listener
> ------------------------------------------------
>
> Key: OAK-1084
> URL: https://issues.apache.org/jira/browse/OAK-1084
> Project: Jackrabbit Oak
> Issue Type: Wish
> Components: jcr
> Reporter: Chetan Mehrotra
> Priority: Minor
>
> Currently Oak dispatches the observation events through a single thread. As
> best practice its recommended that ObservationListener should not block the
> main event thread and instead process the event asynchronously in separate
> thread.
> However it can happen that some rogue implementation does not adhere to this
> best practice and blocks the main event thread. It would be helpful to have
> some support in Oak
> # To identify such listeners which take more time then a defined interval
> # Blacklist such listeners from further event processing and log a warning
> about the same
> This feature would be similar to one supported by Felix EventAdmin [1] where
> EventHandlers are blacklisted in case they timeout
> [1]
> http://felix.apache.org/site/apache-felix-event-admin.html#ApacheFelixEventAdmin-Configuration
--
This message was sent by Atlassian JIRA
(v6.1#6144)
