I have a small single-node machine running TICK (latest version), and receiving a stream of ~50 variables. I have tried both the in-memory and tsi1 indexing methods. I have changed the default retention policy to 5000d. Unfortunately Chronograf only shows about 3 days worth of data. About 1% of disk space is consumed. All journalctl entries appear to be reporting the Chronograf queries (and there are several per second on average).
Any suggestions about what might be wrong? Any logs I should be checking?