Cloudera Manager Service Monitor
Learn about the function and configuration options related to the Cloudera Manager Service Monitor (SMON) and its handling of YARN container usage aggregation.
SMON collects YARN container usage metrics and temporarily stores the raw data in the /tmp/cmYarnContainerMetrics directory.
The following
key processes
are
valid for SMON:
- Collection – SMON gathers container metrics from running YARN containers.
- Storage – The raw metrics are written to the /tmp/cmYarnContainerMetrics directory with updates occurring approximately every hour.
- Aggregation – An Aggregation Job within SMON runs roughly once per hour to consolidate the raw data collected during the preceding hour.
- Cleanup (Default) – After the Aggregation Job completes, the raw metrics files in the /tmp/cmYarnContainerMetrics directory are removed by default.
The cleanup behavior is controlled by the firehose.yarn.retain.raw.usage.metrics.dir.after.job parameter using the following values:
false– The default behavior, raw metrics are purged after aggregation.true– Retention behavior, SMON is instructed to preserve the raw metrics files in the directory.
This parameter can be configured through Cloudera Manager by navigating to .
