Julio C Leal

Hi @ptodev We're having the same problem here. Seems insane, but I think it is related to samples limit exceeded. Look to both graphs   Look at, when the...

> Edit: This can happen any time a scrape is rolled back, not just when the sample limit is exceeded. @rfratto Is there a way to fix this without change...

Yes, agree. Thank you so much @rfratto

> We're running Prometheus in stateless agent mode, and I believe we are seeing this problem pretty significantly. We do make heavy use of sampleLimit settings to manage our metric...

The same here... Running a debian and emacs 29.1 When I dowgrade to 28.1 the errors change

@hagen1778, the part of the doc that I understood that, to run in HA I need has deduplication enabled is [here](https://docs.victoriametrics.com/vmagent/#high-availability), particularly in this sentense > When running HA pairs,...

This means I would need two VMAgent components: one for scraping (handling retries, etc.) and a second dedicated to deduplication and forwarding data to VMInsert. Did I understand this correctly?...

> With HPA you'd have to configure scraping vmagents to shard data by chosen labels across replicas of deduplicators. Which sounds non-trivial. Do you have HPA for scrapers as well?...

Hi @hagen1778 , sorry for delay in answer you. I was traveling last week. 1. The current list of cmd-line flags configured on vmagent. ``` - --promscrape.suppressScrapeErrors=true - --promscrape.suppressDuplicateScrapeTargetErrors=true -...

Hi @valyala , thank you so much for your answer. Yes, the prometheus and vmagent uses the same config file (I use a jinja template and only use if and...