optimal MetricsCacheTimeoutMinutes?

[donsizemore]

This morning I happened to check our Dataverse instance on my phone while it was in mid-metrics update. The homepage was fairly slow 6-8 sec. to load, but was fine once I got logged in. server.log has a ton of these entries:

[#|2019-06-09T06:07:47.969-0400|INFO|glassfish 4.1|edu.harvard.iq.dataverse.util.SystemConfig|_ThreadID=50;_ThreadName=jk-connector(4);_TimeMillis=1560074867969;_LevelValue=800;| Returning 10080 for :MetricsCacheTimeoutMinutes because value must be an integer greater than zero, not "null".|#]

I see that 10080 is one week in minutes and that previous months' caches are preserved, but I'm wondering how this one week cache timeout jibes with our running the metrics update script daily, and whether a longer cache timeout would help as we step through the weeks in each month.

...or whether Odum just needs a bigger box (soon to happen after the DCM)