Hey, so I got this error and did what the doc files suggested, setting max-series-per-database limit exceeded = 0 for unlimited number of series per database.
This resolved this issue, but my question is, does that now consume alot more disk space on the instance that holds the influxdb? What are the retention policies regarding that? Just want to understand this a bit better, thank you
@Tyler_D,
Thanks for your question. The max-series-per-database limit exists only in the inmem index (the default in 1.x). It exists because the inmem index can cause the server to OOM if too many series are added to the index. Since it’s rebuilt on startup, and OOMing server can continue to OOM when it restarts, making it operationally difficult to sort out the cardinality problem. The limit was determined based on general query performance at the limit along with heap usage with typical HW. You might want to look into enabling TSI if you haven’t already and you’re dealing with that type of cardinality (it’s the default in 1.7.x +).
Just to clarify,
lets say I keep the mac-series-per-database set to 0, as mentioned in the grafana docs, what would happen down the road. Would it eventually consume and affect disk usage?
With regards to what you said, as I am new, OOM and HW mean what?
It could potentially affect mem. But you need REALLY high cardinality for that. Most OSS users don’t have any problem (esp after enabling TSI).
