Configure Ops Manager to Monitor Large Deployments

Allocate more RAM to Ops Manager Hosts

1. Open /opt/mongodb/mms/conf/mms.conf in your preferred text editor.

2. Find a line where the JAVA_MMS_UI_OPTS variable defines the -Xms and -Xmx parameters.

3. Update the -Xms and -Xmx values from the default of 8096 MB to more if the hosts have extra free memory.

   Note

   Don't Set Young Generation's Heap Size Explicitly Using -Xmn.

   Starting with Ops Manager 7.0, Ops Manager adopted a Garbage-First (G1) garbage collector method for the JVM. We recommend that you don't set the young generation's heap size explicitly to any fixed value using -Xmn. Setting a fixed value with -Xmn interferes with the G1 garbage collector's default behavior. If you set this value, this prevents the G1 garbage collector from expanding and contracting the young generation space as needed, which causes the collector to stop respecting the pause time value for garbage collections.

   You can set the values for Xmx and Xms to be the same, and retain the pause at its default value: -XX:MaxGCPauseMillis=200.

To learn about how to resolve this error in greater detail, see our troubleshooting entry on the OutOfMemoryError

Improve Monitoring Ingestion

Update app settings on all Ops Manager hosts to improve monitoring ingestion:

1. Open /opt/mongodb/mms/conf/conf-mms.properties in your preferred text editor.

2. Add the following properties:

   # Improved monitoring ingestion
   ingestion.thread.count=64
   ping.thread.count=32
   mms.monitoring.agentlog.writers=16
   mms.monitoring.ingestion.databasenames.poolsize=16
   mms.monitoring.ingestion.diskpartition.poolsize=16
   mms.monitoring.ingestion.groupupdate.poolsize=8
   mms.monitoring.ingestion.hostping.poolsize=32
   mms.monitoring.ingestion.hostversionchange.poolsize=8
   mms.monitoring.ingestion.ipaddresschange.poolsize=8
   mms.monitoring.ingestion.hostlogs.poolsize=16
   mms.monitoring.ingestion.newrelic.poolsize=16
   mms.monitoring.ingestion.profiler.poolsize=16
   mms.monitoring.ingestion.shardchunkcount.poolsize=16
   mms.monitoring.ingestion.totalstoragecache.poolsize=16
   mms.monitoring.rrd.rawdbperf.thread.count=16
   mms.monitoring.rrd.rawdbperf.batch.size=1000
   mms.monitoring.rrd.rawdbstorage.thread.count=16
   mms.monitoring.rrd.rawdbstorage.batch.size=1000
   mms.monitoring.rrd.rawdisk.thread.count=32
   mms.monitoring.rrd.rawdisk.batch.size=1000
   mms.monitoring.rrd.rawhost.thread.count=32
   mms.monitoring.rrd.rawhost.batch.size=1000
   mms.monitoring.rrd.rawnamespace.thread.count=16
   mms.monitoring.rrd.rawnamespace.batch.size=1000
   mms.monitoring.rrd.rawprocess.thread.count=32
   mms.monitoring.rrd.rawprocess.batch.size=1000
   mms.monitoring.rrd.rawsystem.thread.count=32
   mms.monitoring.rrd.rawsystem.batch.size=1000

3. Save the properties file.

4. Restart Ops Manager.