Configure Metric Retention

By default, OpenTSDB stores two weeks of metrics. Based on your requirements, you can change the metric-retention period.

The following cron job runs each day to purge metrics based on the retention period:
$min $hour * * *      $OTSDB_HOME/bin/tsdb_cluster_mgmt.sh -purgeData -retentionPeriod '2 weeks ago'
To edit the metric-retention period:
  1. Depending on the operating system, use crontab -e as the cluster admin (typically mapr) user, or open a crontab file:
    • For RHEL:
      "/var/spool/cron/$MAPR_USER"
    • For SLES:
      "/var/spool/cron/tabs/$MAPR_USER"
    • For Ubuntu:
      "/var/spool/cron/crontabs/$MAPR_USER"
  2. In the following line, update the value of '2 months ago' to the new retention period. You can use any format accepted by the date -- date= command, such as '2 months ago' or '2 days ago'. For example:
    27 5 * * *      /opt/mapr/opentsdb/opentsdb-2.4.1/bin/tsdb_cluster_mgmt.sh -purgeData -retentionPeriod '2 months ago'
    To delete metrics from 1/1/2000 to [current date - 2 days]:
    27 5 * * *      /opt/mapr/opentsdb/opentsdb-2.4.1/bin/tsdb_cluster_mgmt.sh -purgeData -retentionPeriod '2 days ago'
WARNING
Data Fabric monitoring uses 2 MB of disk space per minute per node when HPE Data Fabric Streams metrics are enabled. This is approximately 3 GB per day on a single node or 7 GB per node per day with 3X replication. This streams metrics data is automatically deleted every three days.

If openTSDB is down for more than three days, you may lose the metric data.

NOTE
For more information, see the OpenTSDB scan command documentation.