Roman Khavronenko

Hi @tamcore! If we extend `notifier.config.static_configs` to support auth settings per target - would it suffice?

I think it should be quite easy to implement. However, it won't be compatible with Prometheus. Let's see what we can do.

Hi @torbenaa! > Maybe I should reduce the dedup.minScrapeInterval even further ? It is better to have consistent scrape interval across all jobs. Here's [why](https://www.robustperception.io/keep-it-simple-scrape_interval-id/). If for some jobs you...

> So - instead of me removing the "prometheus_replica" label on ingress and hoping my deduplication interval does not break anything, you could use the labelling as described to better...

Hi @Vladikamira! Thanks for detailed report! Have you tried removing this specific storage node from vminsert's listing just to remove ingestion load? How would storage behave in such conditions?

Just to clarify, the problematic storage that was removed all this time was performing very poorly so almost all the data were re-routed from it. So even removing it from...

Thanks for detailed info! The p.2 screenshots are missing Rerouting but, I guess, I know what it would show. So the removed storage node went calm once the ingestion went...

> for this particular cluster for the whole cluster or for this specific vmstorage? The whole cluster shows good TSID cache hit ratio from previous screenshots ([100%](https://user-images.githubusercontent.com/5423271/184633089-788412fc-0f1a-4c4e-8682-ce2d097dd12d.png)). But from your...

> TSID cache size on restart Cache structure in VictoriaMetrics works in different modes. One of the modes (picked mode depends on amount of occupied memory by the entries in...

> we have 3 overloaded shards and 3 are almost empty "We recommend to run a cluster with big number of small vmstorage nodes instead of a cluster with small...