[
https://issues.apache.org/jira/browse/SOLR-18623?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Houston Putman updated SOLR-18623:
----------------------------------
Affects Version/s: 9.8
(was: 9.9)
> Race-condition in ConnectionManager
> -----------------------------------
>
> Key: SOLR-18623
> URL: https://issues.apache.org/jira/browse/SOLR-18623
> Project: Solr
> Issue Type: Task
> Affects Versions: 9.8
> Reporter: Houston Putman
> Assignee: Houston Putman
> Priority: Major
>
> SOLR-17405 moved SyncConnected and Disconnected off the single-threaded
> callback executor and onto the unbounded pool. That is what lets them be
> applied out of order. Before it, they were ordered. Therefore now it's
> possible to get to a state in Solr where the Zookeeper client thinks that it
> has disconnected and not reconnected.
> The solution is to make sure that the events that the ConnectionManager gets
> are versioned (timestamped) and we only accept events > than the last event
> seen.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
