[ 
https://issues.apache.org/jira/browse/SOLR-17458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18035441#comment-18035441
 ] 

Matthew Biscocho commented on SOLR-17458:
-----------------------------------------

Thanks for finding this Chris. I was not able to reproduce this myself with 
those existing tests but I was able to reproduce it a different way. I spun up 
a thread pool and just hammered CoreContainer initialization and closes on each 
thread and running into the error would hit occasionally as those flaky tests 
show.

I think what is happening is a recording for JVM metrics should be starting on 
CoreContainer initialization but in the OpenTelemetry library that we are 
using, it starts that recording on a separate thread but in the same test we 
initialize and close CoreContainer multiple times. Occasionally we end up 
hitting a race condition in which the main thread calls close() before that new 
daemon thread starts() causing the error. In general, I had JVM metrics 
disabled on all tests except ones that used them for SolrTestCaseJ4 but this 
test bypassed that in that it extended SolrTestCase directly. We can disable 
JVM metrics here as well since this test doesn't use them.

> Metrics: switch from DropWizard to OpenTelemetry
> ------------------------------------------------
>
>                 Key: SOLR-17458
>                 URL: https://issues.apache.org/jira/browse/SOLR-17458
>             Project: Solr
>          Issue Type: Improvement
>            Reporter: Matthew Biscocho
>            Assignee: Matthew Biscocho
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 10.0
>
>          Time Spent: 3h 20m
>  Remaining Estimate: 0h
>
> Solr currently captures metrics with Dropwizard 4. There was some limitations 
> to Dropwizard, biggest one being metrics without tags/attributes making 
> aggregation difficult and requires the Prometheus Exporter to work with 
> Grafana.
> Creating this to track and explore integrating OpenTelemetry into Solr and 
> possibly replace Dropwizard giving a larger exposure of observability tools.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to