Hi Don, This is a master node running services like Ambari, Zookeeper, JournalNode, Hbase Master, HiveServer2, Hive Metastore, Infra Solr, Ambari Metrics Collector, Spark2 History Server, RangerKMS, etc. (So many services but it is a very big and powerful server).
I ended up noticing there were a lot of full GCs for the logfeeder and I have increased the min and max heap from 1 GB to 4 GB for both. The logfeeder has been running with this new heap setting for 3 days now and no issues. Before within 5 minutes of starting logfeeder on that host the CPU would spike. We haven't seen a single CPU spike since. Thanks, Chad On Sat, Mar 23, 2019 at 1:15 PM Don Bosco Durai <[email protected]> wrote: > Hi Chad > > > > Can you provide additional information? E.g. > > > > 1. What services are running on this node > 2. Are there any errors in the logfeeder log file? > > > > Thanks > > > > Bosco > > > > > > *From: *Chad Woodhead <[email protected]> > *Reply-To: *<[email protected]> > *Date: *Thursday, March 21, 2019 at 7:37 PM > *To: *<[email protected]> > *Subject: *Logfeeder high CPU usage > > > > I am running Ambari 2.5.1.0 and I am using LogSearch. I have noticed > logfeeder has started using a ton of CPU on my servers, especially one > server in particular compared to others. Anyone have any experience with > logfeeders using a lot of CPU? > > > > Thanks, > > Chad >
