frozenTimePeriodInSecs and indexer disk space

[khhenderson]

In indexes.conf on the indexer I have my frozenTimePeriodInSecs set to 15552000 = 180 days on indexes that I knew would be big. That seems to be more than I need to keep. I would like to cut it in half 7776000 seconds = 90 days.

Once I do so and restart splunk, is there any additional steps I'll need to do to free up disk space in the indexer?

[jtrucks]

You could just limit the max size of each index by setting maxTotalDataSizeMB and then you don't have to worry about retention length as the retention will be controlled by disk space. You can then set frozenTimePeriodInSecs to your max desired retention time. This has the nice effect of not keeping data longer than you want to have it available (to meet log retention rules for example) but still allow for less than that time period if you have less space available on that indexer.

[khhenderson]

I've look at the documentation and it looks like I can use both maxTotalDataSizeMB and frozenTimePeriodInSecs to keep things in control.

[kristian_kolb]

A good solution if you're short on disk space - but not necessarily a great idea if you have compliance requirements to store some data for a certain minimum amount of time. A sudden surge of events could then reduce your index to only hold a few days (or even hours) worth of events.

/K