Deduplication of data about helm-charts

[Yalafeg]

What I did is that I installed kube-prometheus-stack with a remote_write to victoriametrics using the appropriate URL. What I am afraid of is that the metrics collected with Prometheus will be deduplicated several times at the storage level which will decrease the available capacity. Can you clarify the idea of metrics' collection using Prometheus and saving it to the storage?

And a question about data retention on the storage level, can we do the retention based on the number of GBs occupied and not based on time retention?

Thanks a lot

[tenmozes]

hello @Yalafeg
by default data is not replicated and each pod contains own unique dataset

• replicationFactor > 1 defines how many times you want to duplicate data across nodes. E.g. for replicationFactor=3 vminsert writes data to 3 storage nodes
• dedup.minScrapeInterval, sometime you need to write duplicated data from outside to cluster (e.g. vmagent HA setup) for that you can remove duplication by using dedup.minScrapeInterval. Please take into account - if you set replicationFactor > 1 you need to specify dedup.minScrapeInterval =1ms (https://victoriametrics.github.io/Cluster-VictoriaMetrics.html#replication-and-data-safety)

[tenmozes]

@Yalafeg retention is in Months https://victoriametrics.github.io/#retention

if Prometheus scrapes unique subset of targets the data will NOT be duplicated

[Yalafeg]

Thank you a lot for the details