public class BigdataGangliaService extends GangliaService
GangliaService.serviceNamenamespace. The interval for that counter should be configured separately. The main purpose is to allow applications to decide that some service is missing. TODO Research how to make ganglia recognize a value which is not being reported as "not available" rather than just painting the last reported value. Tmax? DMax? TODO Can metrics be declared which automatically collect history from the sampled counters? It would be nice to abstract that stuff out of bigdata. TODO We should be reporting out the CPU context switches and interrupts per second data from vmstat. Ganglia does not collect this stuff and it provides interesting insight into the CPU workload and instruction stalls, especially when correlated with the application workload (load, vs closure, vs query).
|Constructor and Description|
|Modifier and Type||Method and Description|
addMetricCollector, gatherMetrics, getCanonicalHostName, getDefaultHostReportOn, getGangliaState, getHostName, getHostReport, getHostReport, getHostReport, getMetadataFactory, isListening, isQuietPeriod, removeMetricCollector, run, sendMessage, setMetric, uptime
public BigdataGangliaService(String hostName, String serviceName, InetSocketAddress metricsServers, InetAddress listenGroup, int listenPort, boolean listen, boolean report, boolean mock, int quietPeriod, int initialDelay, int heartbeatInterval, int monitoringInterval, int globalDMax, IGangliaMetadataFactory metadataFactory)
Copyright © 2006–2019 SYSTAP, LLC DBA Blazegraph. All rights reserved.