Shalin Shekhar Mangar updated SOLR-10714:
    Attachment: SOLR-10714.patch

* OverseerTriggerThread reads the /autoscaling.json znode and set a watch for 
future changes.
* Added a TriggerIntegrationTest.testContinueTriggersOnOverseerRestart to test 
this behavior
* Fixed a race condition in OverseerTriggerThread which caused unnecessary 
* OverseerTriggerThread checks that it has not been closed before attempting to 
wait for the condition
* Sometimes the coreContainer.getZkController can return null because of the 
way overseer is created (in ZkController's constructor) which causes the 
trigger to NPE in the constructor -- I'll open a separate issue about this 
because this isn't related to autoscaling as such. To workaround this problem 
TriggerIntegrationTest ensures at least two nodes be available before running 

> OverseerTriggerThread does not start triggers on overseer start until 
> autoscaling config watcher is fired
> ---------------------------------------------------------------------------------------------------------
>                 Key: SOLR-10714
>                 URL: https://issues.apache.org/jira/browse/SOLR-10714
>             Project: Solr
>          Issue Type: Sub-task
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: SolrCloud
>            Reporter: Shalin Shekhar Mangar
>            Assignee: Shalin Shekhar Mangar
>              Labels: autoscaling
>             Fix For: master (7.0)
>         Attachments: SOLR-10714.patch
> Thanks to [~ab] for catching this. The OverseerTriggerThread only sets a 
> watch but doesn't read trigger information by itself. Therefore no triggers 
> are started on overseer restart until the autoscaling zk node changes and the 
> watcher is fired.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to