I am still super happy with InfluxDB v2. I like the Flux query language a lot.
Regarding HA: I have configured telegraf to provide an InfluxDB listener. From that entry point telegraf writes all the received data out to three endpoints (local InfluxDB v2 instance with 90 days retention, InfluxDB Cloud free of charge instance with 30 days retention and a local MQTT broker). Of course, the listener is write-only. As I do all my analytics and automation tasks in Grafana or InfluxDB Tasks this is okay for me.
telegraf buffers the data if one of the endpoints is not available. Once I comes back online, the buffered data is flushed to it. This way one of the two InfluxDB instances can go down and I can still work with the other. Kind of ‘poor man´s HA’. And I never lost any data.