Newbie _internal stats question



I have a database that I created using a default config. When I look at the diskBytes for the filestore I see it oscillating between zero and some value:

> SELECT "diskBytes" FROM "tsm1_filestore" WHERE ("database" = 'metrics') and time > now() - 1m
name: tsm1_filestore
time			diskBytes
1541386880000000000	62865152
1541386880000000000	0
1541386890000000000	62865152
1541386890000000000	0
1541386900000000000	62865152
1541386900000000000	0
1541386910000000000	62865152
1541386910000000000	0
1541386920000000000	62865152
1541386920000000000	0
1541386930000000000	62865152
1541386930000000000	0

… this behaviour started recently, when diskBytes hit 63.7 million. There are no repeated compactions in the log. I see this behaviour paralleled by tsm1.memBytes, where the one has the opposite gradient to the other, but the magnitudes are different (obviously).

I’m new to Influx, what’s happening here? Is this simply a product of the way the stats are generated, where intermediate points are zero, or is this reflecting what is actually going on?

Note: there are no deletes being sent to influxdb. The number of measurements only increases.