#InfluxDB v3.0+ vs Prometheus?

1 messages · Page 1 of 1 (latest)

undone moss
#

Does anyone have a preference here? I was going to swap to Promtheus because I've found v2.7 to be challenging when you want to create a db backup, but now that v3.0+ is available with SQL I am wondering what experience others have had.

#

TBH it doesn't look like the homeassistant integration supports v3.0+ yet, so I guess it's a moot point

south coral
#

I switched from prometheus to VictoriaMetrics because the long term storage of prometheus was somewhat inefficient and it used more resources. I find the syntax of InfluxDB hard to understand too. V3 seems to have some limitations. What's the issue with v2? The community support for Influx will be better as it's just more popular.

undone moss
#

I recently had a bad experience where I was trying to dump/back-up a year's worth of old data, and it was very challenging.

  • There wasn't a black and white, "backup" method baked into the database, so it had to run via command line
  • Creating a backup required the admin token, and there wasn't an easy path for recovery
#

What are the limitations with V3? I heard that it locks you out of querying data that's more than a year old, but that was on reddit

south coral
#

Yeah that. And flux might be gone too. If you run it as addon the HA backup system includes it.

undone moss
#

We're running them both in separate containers, so no-dice on the HA backup system

south coral
#

You have the addon tag here though.

undone moss
#

Ohhhhhh my bad

#

We use the integration to push data into V2

south coral
#

Hmm. If you virtualize with something such as Proxmox VE you could just back up the whole VM or LXC container. I don't think prometheus has a better backup system either.

undone moss
#

That's good to know; I appreciate you taking the time to help

#

You're the second person I've seen mentioning VictoriaMetrics, so maybe that's something we'll look into

tacit pike
#

I just use a simple backup script for influxdb v1.8, is v2 different in that regards?

undone moss
#

No, I don't think so, backups in V2 require that I provide the API key and a couple CLI commands; so it's not super complicated, just harder to do since it runs out of a container

tacit pike
#

Ah ok. I just mount a second /backup volume in the container, and run the backup command on a schedule.

undone moss
#

That's a clever solution 👀

tacit pike
#

The backup is just then (I'm using docker compose):

docker compose -f /srv/apps/docker-compose.yml \
  exec -T influxdb sh -c '
  rm -f /backup/* && \
  influxd backup -portable /backup
  '

The data is then available on the host. The backup software actually runs the command before backing up the host.