Symptoms
The following is observed when monitoring capacity of Data Domain from the Avamar Administrator dashboard.The forecast information changes erratically over the course of a week rather than tracking the longer term capacity trend of Data Domain.The dashboard shows an oscillation between a shorter forecast (for example, 3 months) followed by a much longer forecast (eg. 3 years).The longer forecast aligns with the weekly 'cleaning' maintenance. By default, this occurs on Wednesday.The screenshot below shows that the forecast for Data Domain #2 is ~3 years. This forecast disagrees with the projected trend found in the Avamar Enterprise Manager capacity. It also disagrees with the Data Domain Enterprise Manager "File System - Consumption" graph.Both graphs agree that Data Domain #2 has around 3 months of remaining capacity following the current trend based on the weekly peaks. These occur before DD cleaning.
Data Domain Enterprise Manager graph is below.
Cause
This is due to the sensitivity with which the Dashboard forecast figure is generated.
Resolution
Since Avamar and Data Domain cleaning cycles differ, the Avamar dashboard forecast algorithm may need tuning to take account of this to produce more accurate results which reflect longer term capacity trends.
Workaround:1) Tune the mcserver.xml parameter.a) Increase the value for the "capForecastDataDays" parameter in the /usr/local/avamar/var/mc/server_data/prefs/mcserver.xml file.b) Restart the MCS service with the following commands (or see the system Administrator Guide).dpnctl stop mcsdpnctl start mcs2) An alternative workaround would be to make use of the Data Domain Enterprise manager 'Space Prediction' comment at the base of the "File System - Consumption" graph shown earlier.
