Influxdb 1.8.x - Problems in the occupation of the datadir

Hi, everyone!

I’ve been having problems for a few days in the influx datadir.
The influx datadir has 106G, dedicated just for it, the folders (data, wal and meta) are all there and I’ve already occupied more than half of it - an abnormal behavior.
After restarting the service, the occupation of the datadir returns to ~20%.

The machine’s resources are ok.
I’ve a VM - CentOS 7, 24vCpus and 30Gb RAM.

I have an Influx v.1.8.4 with 523 measurements and 3 fields (time, id and value) each.

My retention policy:
name duration shardGroupDuration replicaN default


autogen 17544h0m0s 720h0m0s 1 true

Influx_inspect:

influx_inspect verify -dir /var/influxdb/
Broken Blocks: 0 / 38334621, in 56.972801362s

The only error I found was this:

influx_inspect report /var/influxdb/data
report: not a valid shard dir: /var/influxdb/data/flex/autogen/100/1.tmp

I didn’t find references about it.

I think the compression is having problems.

Any suggestion?

Best regards!

When the compression fails you should find some related entry in the influxDB log.

I had a similar problem twice in two different (windows) environments but after a server/service restart it was gone… and never happened again

Updating…

I’ve a backup’s routine and works at midnight, I removed from crontab.
Today, the command influx_inspect report /var/influxdb/data was executed successfully and growth is stable.
My backup’s command:

influxd backup -portable -database $DATABASE $REPOSITORIO/$DATA/ > $LOG/EXECUTION.log

I have other influxdb servers with the same routines but with different versions of influxdb (< 1.8.4).

Updating

I will update to version 1.8.6.
I haven’t identified these issues in this release.