I’m a bit of a noob atm so please excuse any terminology failings!
I’ve been running an instance of inluxdb for my grafana instance and had started on influxdb 1.24 with the admin panel (1.20) I recently upgraded to current version 1.72 hoping that any optimizations may help me get out this hole but and am still having the same issue where i try and delete a series of data from the database the database goes into an endless cycle at 100% cpu load and keeps churning through trying to delete the data. Since the upgrade at least the churning has stopped!
My database is around 400gb and when i try and remove this particular series the database consumes a further 100gb to take it to 500gb and then falls back to 400gb in an infinite loop.
The series in particular is simply all values to do with a particular linux host. This particular host was a docker instance that had decided to restart itself every 30 seconds and kept making unique “veth-xxx” instances of the NIC every time it was restarted and logging the data to influx.
I believe it probably hit a 10,000 value limit or similar and influx has been tracking all of these values since.
How can i drop the data for the host in question?
I’ve been using the admin panel with the following query.
DROP SERIES WHERE host='myservername'
I’ve even tried the following query which seems to work as i can show the data from it.
DROP TAG VALUES FROM net WITH KEY IN ("interface") WHERE host = 'myservername' AND "interface" =~ /veth/
This never seems to finish.
When i try and get more granular and delete a particular instance it completes with success but the data dosn’t remove!
DROP SERIES WHERE host = 'myservername' AND "interface" = "vethfff429d"
Can someone help me out? It seems that this should be quite straight forward and i’m assuming the sheer amount of measurements that it’s trying to drop is the issue.