action #94492
Updated by mkittler about 3 years ago
## Motivation Currently the default retention period is in-place: ``` > use telegraf Using database telegraf > show retention policies name duration shardGroupDuration replicaN default ---- -------- ------------------ -------- ------- autogen 0s 168h0m0s 1 true ``` According to the documentation that means we store the data infinitely¹. It would be possible to configure a time-based retention policy per database (we store all our monitoring data in one database called `telegraf`): https://docs.influxdata.com/influxdb/v1.8/query_language/manage-database/#create-retention-policies-with-create-retention-policy This raises the question which durations we'd like to configure. It would also make sense to "[downsample and retain data](https://docs.influxdata.com/influxdb/v1.8/guides/downsample_and_retain/)" to store data with less granularity over time. I haven't found anything about disk-space-based retention policies within the documentation of InfluxDB so I **assume** this is not an option. --- ¹ > When you create a database, InfluxDB creates a retention policy called autogen with an infinite duration… ## Acceptance Criteria * **AC1:** **AC 1:** The directory monitor.qa.suse.de:/var/lib/influxdb has significantly less than the current 120GB * **AC2:** Grafana is able to show at least 6 months worth of (optionally downsampled) data ## Suggestions * Ensure we have a backup before the process remove data * Check what the current InfluxDB version 1.8 provides * If we have downsampling, we should aim for at least 1 month of full-sampled data data * Checkout previous attempts to export data * dump is still lying around at `/var/lib/influxdb/dumps/postgresql`; it can be deleted if the disk space is needed * see #93922#note-10 but do *not* make the same mistake of using `grep` again for filtering