[
https://issues.apache.org/jira/browse/SOLR-14918?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17209474#comment-17209474
]
Jan Høydahl commented on SOLR-14918:
------------------------------------
I cannot see that this is a common scenario. The node is healthy. Various
application level requirements can probably best be enforced with custom logic.
I understand that for your particular application you want nodes without a core
to be killed by AWS. Can you perhaps create a custom health check plugin that
delegates to /api/node/health for health checks? Or perhaps a sidecar
application running alongside Solr which embeds this logic?
> add "failIfEmptyCores" option for HealthCheckHandler
> ----------------------------------------------------
>
> Key: SOLR-14918
> URL: https://issues.apache.org/jira/browse/SOLR-14918
> Project: Solr
> Issue Type: New Feature
> Security Level: Public(Default Security Level. Issues are Public)
> Affects Versions: 8.6
> Reporter: Taisuke Miyazaki
> Priority: Major
> Labels: features, ready-to-commit
> Time Spent: 10m
> Remaining Estimate: 0h
>
> h2. Background
> We are hosting Solr on AWS EC2.
> We have set up Solr to start in systemd and create a core at startup. (The
> replica type is "pull".)
> I manage those EC2 instances in the AWS AutoScaling Group and use TargetGroup
> to access Solr via the Application Load Balancer.
> For health checks in TargetGroup, we use the requireHealthyCores option using
> /api/node/health (henceforth called HealthCheckHandler).
> In most cases this is not a problem.
> However, this can cause additional problems with a rare core creation failure
> issue at startup.
> If you use the requireHealthyCores option for HealtchCheckHandler, the status
> will be OK if the core is not there in the first place, so the health check
> will be OK even though the core is not present (an unusual condition).
> h2. Proposal
> Therefore, we propose to add an option to set the status to Fail if the core
> does not exist. That way, we can make the health check fail even if the core
> was not created for some reason.
> We believe this option should only work if the requireHealthyCores option is
> true.
>
> Translated with www.DeepL.com/Translator (free version)
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]