Robert Fratto

@nerdblogpl Pushing directly from an application to a DB removes the ability to have centralized transforms and risks data loss if your app crashes. Because of this, you normally want a collector in between your app and the database anyway, regardless of whether you're using push/pull.

@nerdblogpl I used to work on Alloy, so I can answer: pulling metrics has a few benefits, like being able to tell you if target is down (the up metric).

I don't understand the idea behind pull-based metrics systems like LGTM stack. All I can see is "you don't push metrics, it scrapes your data! weeell, except you need a separate Grafana Alloy process to totally-not-push-bro the data on everything that can't expose extra ports".

@after_ephemera @grafana i don't have any good ideas for how to handle that today, but i also want something like this. i'm planning on adding on YAML support within the next few months that uses Alloy syntax expressions embedded inside ${} stanzas. sorry it won't be TOML, which i prefer over YAML tbh

@robertfratto @grafana shame. i actually don't mind it that much, i just wish there were an easier way to templatize it (deploying n services to cloud run with alloy sidecar containers, want to have n target labels on them). any suggestions on a good way to do this?

if you're down today remember that someone at @grafana probably got a promotion to staff by writing a bespoke alloy dsl and parser instead of just... you know, using toml github.com/grafana/alloy/…

@rtl_sdr Glad to hear you got it working!

@robertfratto Thank you for telling me how to debug and Alloy's UI. Solved it! Temperature and humidity were visualized.

Grafana Alloyはドキュメント通りに設定してもexporter拾えないしまだまだダメだなぁ…残念 grafana.com/docs/alloy/lat…

@rtl_sdr Hello! I agree that this is confusing. Currently, scrape errors are only logged when debug log lines are enabled. Alternatively, you can Alloy's UI[1] to debug your prometheus.scrape component to see whether targets are being scraped successfully. [1]: grafana.com/docs/alloy/lat…

ポート番号ミスってた…😭 何もエラーログ出さないのヤバすぎる prometheus.scrape "sht3x_exporter" { targets = [{ __address__ = "127.0.0.1:2112", }] forward_to = [prometheus.remote_write.metrics_service.receiver] scrape_interval = "10s" }

Alloy is the spiritual successor to Grafana Agent, blending the best of both the OpenTelemetry and Prometheus! 💪

@kkakkoyun @sh0rez sorry about the unfortunate timing here. feel free to reach out to me personally or publicly if you have feedback!

@robertfratto @sh0rez Thanks for the link. I'm already digging through those. Happy to share my feedback 🤗 The frustrating part for me is I have just moved everything to agent flow, and I have to do this again😩

"Grafana Agent, our existing OSS telemetry agent, and Grafana Agent Operator are now deprecated and will enter into Long-Term Support (LTS) beginning today (April 9, 2024)." grafana.com/blog/2024/04/0…

At #GrafanaCON, we've just unveiled Grafana Alloy, our open source distribution of the OpenTelemetry Collector: bit.ly/3J9JkCj Alloy is a telemetry collector that is 100% OTLP compatible and offers native pipelines for OpenTelemetry and #Prometheus telemetry formats, supporting metrics, logs, traces, and profiles.

@kkakkoyun @sh0rez we wrote a series of migration guides here: grafana.com/docs/alloy/lat… I'd love feedback about how we could make this process easier if you find it's not straightforward enough.

@sh0rez Just the amount of unnecessary work that I will have to do for migration 🤷‍♂️ Tell me if I’m missing something. Is it straight forward to migrate?

@pavangudiwada_ @grafana I don't see why it shouldn't be possible, but it sounds like I'll need more info to see what you've tried and identify where it hasn't been working so far, but that might be hard on here. Would you mind opening a discussion in the grafana/agent repo for a longer form chat?

@robertfratto @grafana I'm trying to get the same setup working with Grafana Agent without Kube Prometheus stack. I'm using the flow mode, not intentional, but I think its the default. Is it possible to set this up as I described? I'm only able to get cAdvisor to scrape itself.

Did you work with @grafana agent flow before? Is it possible to get the cAdvisor and kube-state-metrics working with it? All without Prometheus! Almost giving up trying this

@matryer if your semicolon is slowing you down, there’s some great Irritable Semibowel Syndrome medication out there to help relieve your symptoms

I don’t even have time to read tl;dr The semicolon slows me down.

🎁 Just in time for the holidays — Grafana Agent v0.38 is here! Rejoice in new #OpenTelemetry components, configuration improvements, and enhanced log collection. bit.ly/49VZWcR

grafana agent v0.38.0 is now available with over 50 new features, enhancements, and bugfixes! go check it out :) github.com/grafana/agent/…

Launch day! Announcing Tailscale for the Apple TV, new in tvOS 17. Connect directly with devices on your tailnet and pick up new traffic routing superpowers tailscale.com/blog/apple-tv/

BIG NEWS🚨Grafana Pyroscope 1.0 just dropped! 🔥 -✅ Petabyte-scale ingest 📊 -✅ Low Overhead 🙌 -✅ Easy Upgrade 🛠️ #continuousprofiling has never been better! Check it out blog post and upgrade guide! 👉 grafana.com/blog/2023/08/2…

@matryer how do you know when you’ve chewed other food down to the size of rice?

So, do we need to chew rice or not? I chew other food down into the size of rice.

With Grafana Agent v0.35, it's easier than ever to get started and to horizontally scale. Check out all the highlights from the latest release. grafana.com/blog/2023/07/1…

@matryer Thanks for trying Mat, but there’s something in the way of the time lapse.

I made a time lapse for you. #london