I’m using InfluxDB v1.8.10 as it’s provided with my home automation system (openhab 3.1.0) and I’m a big fan of it. I backup the database with a daily cron-job using influxd backup -portable.
Now I found out that during this backup job which runs about 5 mins the memory of my raspberry decreases dramatically and I receive socket timeouts for entries that the openhab home automation writes to influx during this time.
As the home automation system uses ZRAM which I think should not affect the influxd setup but I don’t know how to proove that.
Or is the ZRAM used and as soon as the ram is filled with the backup file and forced to write to physical disk the concurrent write operations cause these timeouts?
Any ideas on how to proceed with any analysis or alternative suggestions are welcome!
try iolimit / renice / taskset on the backup process to constrain it’s reach.
use typical tools, htop etc. etc. to narrow down the bottleneck further.
zram gets you ‘breathing room’ at the cost of a small amount of cpu. in a nutshell, if nothing is maxed then a subprocess is likely holding up more critical stuff. if it’s a synchonous thing you can’t change, then you need to better resource it. if it’s asynchronous then it may be hogging too much resources.
in general tho’ by the sounds of things you’re likely just reaching the hardware limits of the platform so a change in hardware or separate host might be in order.
