* Remove SRL.listConfigDir (unused)
* Remove SRL.getDataDir
* Remove SRL.getCoreName
* Remove SRL.getCoreProperties
XmlConfigFile needs to be passed in the substitutableProperties
IndexSchema needs to be passed in the substitutableProperties
Remove redundant Properties from CoreContainer constructors
* Remove SRL.newAdminHandlerInstance (unused)
* Remove SRL.openSchema and openConfig
* Avoid SRL.getConfigDir
Also harmonized similar initialization logic between DIH Tika processor & ExtractingRequestHandler.
* Ensure SRL.addToClassLoader and reloadLuceneSPI are called at most once
Don't auto-load "lib" in constructor; wrong place for this logic.
* Avoid SRL.getInstancePath
Added SolrCore.getInstancePath instead
Use CoreContainer.getSolrHome instead
NodeConfig should track solrHome separate from SolrResourceLoader
* Simplify some SolrCore constructors
* Move locateSolrHome to new SolrPaths
* Move "User Files" stuff to SolrPaths
On windows, these objects can't be inspected due to security restrictions. So the test runner fails the tests since it does not know how big the leak is.
SOLR-13234 broke prometheus exporter startup from the startup scripts because there was a mismatch between the dependency list in ant/ivy and those actually added to the classpath by the script. This commit changes the script to add all solr dependencies to the classpath.
WhenSolrExporterIntegrationTest.jvmMetrics ran on a JVM with the Turkish locale, (test seed: 62880F3B9F140C89). The JVM metric for terminated thread-count has a dotless-i e.g. termınated.
This causes the check for matching metrics to fail.
We could normalize the text in this case, however I think it's better to ensure we have the correct total number of JVM thread metrics rather than looking at Prometheus labels which maybe localized.
This closes#605.
This changes the prometheus exporter to collect metrics from Solr on a fixed interval controlled by this tool and prevents concurrent collections. This change also improves performance slightly by using the cluster state instead of sending multiple HTTP requests to each node to lookup all the cores.
This closes#571.