[SOLR-3083] all JMX Mbeans are identified as java.lang.String even if numeric

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 4.0-ALPHA
• Fix Version/s: 4.0-ALPHA
• Component/s: SearchComponents - other
• Labels: interoperability, jmx monitoring statistics
• Assignee: Ryan McKinley
• Reporter: Tagged Siteops
• Created: 01/Feb/12 00:05
• Updated: 10/May/13 10:39
• Resolved: 07/May/12 17:38

Description

When trying to monitor our Solr instances with JMXtrans (https://github.com/lookfirst/jmxtrans), we discovered that all the Mbeans were identifying themselves as java.lang.String. This breaks the GangliaWriter in JMXtrans because the software decides it's a string instead of a numeric values and you can't graph a string. Other software we are monitoring with JMXtrans (e.g. Kafka) correctly identify their numeric values as numbers (doubles in this case).

For example, avgTimePerRequest in Solr is identified as a string which makes no sense. This behavior makes Solr less interoperable with the rest of the Java ecosystem.

The problem code is in JmxMonitoredMap.java

NamedList dynamicStats = infoBean.getStatistics();
if (dynamicStats != null) {
for (int i = 0; i < dynamicStats.size(); i++)

{ String name = dynamicStats.getName(i); if (!staticStats.contains(name)) attrInfoList.add(new MBeanAttributeInfo(dynamicStats.getName(i), String.class.getName(), null, true, false, false)); }

I will try to attach a screenshot.

Thanks for Solr! We use the heck out of it, it's awesome.