What is ixdb/base?

[Patrick M. Hausen]

Hi all,

I have a dedicated Ubuntu VM with a 20 G virtual disk for TrueCommand. Running routine updates I noticed the disk to be full.
ixdb/base is taking up 10 G. For an enthusiast installation managing 3 small systems, 2 Core, 1 Scale.

Any advice greatly appreciated.

Thanks,
Patrick

 

[aervin]

That's the TC DB @Patrick M. Hausen . You probably have a TON of historical stats data for your NAS fleet stored there. We exposed a "Stats retention policy" knob in the Administration Settings page. Lowering that might help. Let us know!

 

[Patrick M. Hausen]

I had that set to 24 months. I did not expect log data from three systems to be such a behemoth.

OK, I deleted the apt cache and the systemd journal and got enough wiggle room to actually start TrueCommand. Set it to 1 month. Usage did not go down. The TrueCommand system is 3 months old ... what is taking up 10 friggin' gigs? As I said the "NAS fleet" is three hobbyist systems ...

I might as well just start over and monitor disk usage more closely this time. I was planning to move my docker based applications to my Scale system, eventually. but that has got only 200 G of mirrored pool, because I partly built it from leftover parts. Probably not the best idea. We'll see. For the moment I'll nuke the container and the data and start fresh.

Thanks for your help.

 

[Jaron]

Out of curiosity, How long had that VM been running collecting stats from the 3 systems?

 

[aervin]

That is a lot of storage consumption. By default, TC polls NASs for stat data every 20 seconds. Lowering the stat retention policy and increasing the poll interval should reduce storage consumption overall. For future TC versions, we're looking to drastically reduce the NAS stats that are stored in the DB. Thanks as always for reporting @Patrick M. Hausen

 

[Patrick M. Hausen]

Out of curiosity, How long had that VM been running collecting stats from the 3 systems?

About 2 months. I set it up some time in late August.

 

[Patrick M. Hausen]

That is a lot of storage consumption. By default, TC polls NASs for stat data every 20 seconds. Lowering the stat retention policy and increasing the poll interval should reduce storage consumption overall. For future TC versions, we're looking to drastically reduce the NAS stats that are stored in the DB. Thanks as always for reporting @Patrick M. Hausen

You have heard of RRD, yes?

Seriously, wouldn't a time series database like influx be suited much better than postgresql? My influxdb for Grafana is collecting data since June and is barely a gig in size. 2 FreeNAS systems reporting via the builtin collectd. One OPNsense - same.