Its never been easier to make rl environments: "Make an environment from my local Grafana at 127.0.0.1 Cover dashboards and folders, including creating, renaming and deleting folders." -> link
brandRising
Grafana Labs on X
Grafana's biggest theme is 39 people running it in their own stack who say it does the job, ahead of 9 who love it loudly. The recurring complaint is the manual work of wiring it up, and comparisons are spread thinly across many other tools.
First written 27 Oct · what changed · from 95 posts by 86 people
Tracking posts that say “grafana”
Track this for your marketPosts, last 24h
40
+38%vs. previous 24h
Posts, last 7 days
213
110 in the last 72h
Velocity
30/day
7-day average
Momentum
+31%
Last 24h vs. a normal day this week
Posts per day, last 30 days
as of Nov 3, 09:00 UTCWhat changed
as of 27 Oct · checked once a monthFirst written 27 Oct. When what people say changes, it shows here.
Tone· 58 posts from 54 people· written 27 Oct
How do people feel about Grafana Labs?
Most of the talk is plainly positive: 39 people running Grafana in their own stack say it does the job without fuss. 9 are loud about how much they love it, and only 3 say they gave up.
Plain endorsement comes from people running it themselves: a container on a Raspberry Pi, a home lab holding InfluxDB and MQTT, network flow charts, DORA metrics on an internal developer platform, home power draw checked on a local instance. What they credit is the viewing layer: metrics collected by Prometheus or exported as OpenTelemetry land in Grafana, where they can be seen and alerted on, and they pass "Prometheus + Grafana" on to others as the pairing to use.
The loud ones are 9 people. Dashboards "make complex infrastructure way easier to actually understand"; one built and open-sourced a Grafana plugin after a talk, another calls the panels "such a beautiful beautiful place to live inside of" and credits them with reduced fatigue, and one rates the Grafana Labs AI team.
Not everyone stays. 3 people say it did not stick, over running it with Prometheus and dashboards that need more upkeep than the thing they watch. 3 are still trialling it or running it beside another tool.
- Works well for them43
- Loves it9
- Frustrated or gave up3
- Still weighing it3
See 5 of 58 postsHide the posts
A Simpler Way to Experiment With Optimum Exploring networking infrastructure becomes much more useful when you can observe what's happening instead of relying on assumptions. That's where HOP comes in. Optimum HOP is a Docker Compose framework that helps developers and operators experiment with mump2p and monitor an Optimum Gateway through Prometheus and Grafana. The workflow starts with a running Optimum Gateway. Once Gateway telemetry is enabled, you can configure Prometheus to collect its metrics and launch the monitoring services using the provided Docker Compose configuration. Grafana is available locally at localhost:3000, Prometheus at localhost:9090, and the Gateway API and metrics endpoint at localhost:48123. This setup gives operators a practical way to inspect Gateway health and observe the metrics exposed by their deployment. Instead of treating the Gateway as a black box, they can use monitoring data to better understand its operational state. HOP also provides flexibility. With HOP Lite, you can monitor an existing Gateway. HOP Full offers a broader local environment containing the Execution Layer, Consensus Layer, Gateway, and monitoring components. The important point is that HOP is a testing and observability framework, not a replacement for mump2p or the Gateway itself. For anyone exploring Optimum's infrastructure, it provides a practical starting point for learning, testing, and monitoring. @get_optimum
Had a talk @KochiFOSS today regarding Grafana and how I've used Grafana as an alternative frontend for my side projects. Ended up clauding a plugin for Grafana which made it into a presentation software. Opensourced the plugin at link Grafana ftw🙌
The AI team at Grafana Labs is one of the best agentic application teams in the world imho. I’m playing it cool here, but I’m blown away daily by what they are creating.
Ngl that's probably how it goes, one more dashboard every time something breaks. At some point you're maintaining the monitoring more than the thing it watches. Do you actually run Grafana for anything right now?
5 of the 58 posts behind this answer, one per person. The other 53, and every new one as it’s posted, come with tracking. Get early access
Pain points· 24 posts from 21 people· written 27 Oct
What do people complain about with Grafana Labs?
Manual work is the biggest complaint: 9 people say assembling and keeping Grafana's stack running is fiddly. 6 report bugs, wrong values or outages, and smaller groups raise exposure and late alerts.
Setup is the big one. 9 people describe putting the pieces together by hand: exporters installed one at a time, scrape rules written line by line, dashboards built panel by panel. One says a full Prometheus and Grafana build takes half a day, another shouts that Grafana with Prometheus is too complicated. The defaults miss things too: a short scrape interval lets brief spikes through, log lines arrive as opaque strings that need parsing, and readings of different ages need a moving average to be useful. One describes "maintaining the monitoring more than the thing it watches".
6 report things breaking or misreporting: a query pointed at a Loki service that did not exist, a critical path algorithm implemented incorrectly after copying Jaeger's code, a port charted at zero, an inefficient chart query pegging the CPU of the node running Prometheus.
4 raise exposure — an instance reachable from the internet with Redis behind it, a token scoped and tail sampling added before handing access to an agent, a deployment still running without a password. 2 find alerts arriving late or at bad moments.
- Complex and fiddly to run12
- Bugs, wrong values and outages6
- Security and exposure risk4
- Alerting disappoints2
See 5 of 24 postsHide the posts
服务器哪里出毛病，现在可以直接问 Claude Code 了 服务器半夜告警，SSH 上去敲 top，CPU 不高；换 iostat，磁盘也正常；再翻 dmesg，一屏日志看不出名堂。 等你终于想起来看网络，那一波抖动早过去了。 想上正经监控，Prometheus 加 Grafana 是标准答案，exporter 一个个装，抓取规则一条条写，仪表盘再一个个拼。 一套弄下来小半天，默认 15 秒采一次，几秒钟的尖刺照样漏掉。 Netdata 省掉了这些步骤，一行命令装上，几百个指标自动发现、每秒采集，仪表盘自己长出来，每个指标还单独训练一个异常检测模型。 这两年它又多了一手，内置 MCP server，Claude Code 能直接接上来，在终端里问一句「这台机器刚才怎么了」，它自己去翻指标。
Good ol sysadmin Sunday. Pushed a couple of PRs to my homelab infra code: link - Grafana was querying a Loki service that didn't exist. Fixed - Alertmanager now pushes alerts to a Telegram bot - 3 more servers added to Prometheus + Grafana Also ran memtest on a server that's been acting up. Caught the bad RAM behind weeks of etcd corruption on 1 of my 3 Kubernetes control plane nodes
working on a critical path feature for axiom's tracing fun fact: jaeger and grafana (which copied jaeger's code) implement CRISP (best critical path algorithm from uber) incorrectly so they get tripped up by slight mis-timings
Ngl that's probably how it goes, one more dashboard every time something breaks. At some point you're maintaining the monitoring more than the thing it watches. Do you actually run Grafana for anything right now?
Щось я трошки не врубась як Redis статистику у grafana затягнути. Воно через плагін вміє прямо з нього все показати, але ж у мене Redis не стирчить в інтернет, як вини собі це уявляють? 🙃
5 of the 24 posts behind this answer, one per person. The other 19, and every new one as it’s posted, come with tracking. Get early access
Alternatives· 16 posts from 13 people· written 27 Oct
What are the alternatives to Grafana Labs?
Comparisons are scattered: the largest group is 2 people who name Zabbix, and every other tool comes up from one or two people each. What the rest weigh is a full Prometheus + Grafana build against something quicker to stand up.
The comparisons do not add up to a favourite. 2 people name Zabbix, and one who runs Zabbix 8.0's APM data prefers reading it in Grafana. Every other tool comes from one or two people.
Install effort and cost colour the rest. Netdata is weighed against a Prometheus + Grafana build — one-command install, per-second metrics and built-in anomaly detection against Grafana's ecosystem and query flexibility. StatLite is offered for a single Spring Boot app on a $5 VPS, where "Prometheus + Grafana is usually overkill"; Beszel trades depth for a footprint small enough for many small machines. On logs, Loki is set against Splunk on ingest cost with LogQL as the catch, and VictoriaLogs is called "easier to configure and operate than Loki". One person is steering away from CloudWatch Logs toward Loki; others suggest an all-in-one like OneUptime or a managed backend like Better Stack.
A few weigh Grafana beside tools they already run — Datadog for agent traces, HyperDX while deciding — or came to it from Google Data Studio, for dashboards kept as code.
- Zabbix2
- Netdata2
- StatLite2
- Splunk1
- HyperDX1
- Datadog1
See 5 of 16 postsHide the posts
My first Vietnamese article on Viblo 🇻🇳 Spring Boot monitoring on a small VPS without Prometheus/Grafana, using StatLite. Vietnamese devs: I'd love feedback on the translation! link
服务器哪里出毛病，现在可以直接问 Claude Code 了 服务器半夜告警，SSH 上去敲 top，CPU 不高；换 iostat，磁盘也正常；再翻 dmesg，一屏日志看不出名堂。 等你终于想起来看网络，那一波抖动早过去了。 想上正经监控，Prometheus 加 Grafana 是标准答案，exporter 一个个装，抓取规则一条条写，仪表盘再一个个拼。 一套弄下来小半天，默认 15 秒采一次，几秒钟的尖刺照样漏掉。 Netdata 省掉了这些步骤，一行命令装上，几百个指标自动发现、每秒采集，仪表盘自己长出来，每个指标还单独训练一个异常检测模型。 这两年它又多了一手，内置 MCP server，Claude Code 能直接接上来，在终端里问一句「这台机器刚才怎么了」，它自己去翻指标。
Prometheus + Grafana is usually overkill for one Spring Boot app on a $5 VPS. StatLite turns Actuator into charts for health, traffic, latency, memory, and restarts. One Go binary, SQLite, no agents. link
stop paying for Splunk. seriously. Grafana Loki. $0. it takes one minute. i'll wait. Splunk bills ~$150/GB/day ingested. Loki indexes labels only and writes logs to object storage. AGPL-3.0. catch: LogQL not SPL. one weekend to relearn, then it's yours. overpriced line item?
Zabbix 8.0のAPM、集めたデータを、Grafanaに参照させる方が見やすい
5 of the 16 posts behind this answer, one per person. The other 11, and every new one as it’s posted, come with tracking. Get early access
Questions· 4 posts· written 27 Oct
What do people ask about Grafana Labs?
The 4 questions are all one-offs from different people, asking what Grafana does on its own: whether incidents auto-resolve, whether traces and logs reach an instance they already run, and whether Watchers learns a baseline.
Two of the questions are about getting data in: whether agent traces already flow into an existing Grafana or stay console-only, and whether logs can be pushed straight to a local Grafana instead of a chat channel. One asks how Grafana actually resolves an incident on its own. One asks @grafana directly whether Watchers learns what normal looks like or just follows rules.
Each comes from a different person and stands alone — no question repeats here. They sit in threads about wiring agents, traces and logs into Grafana, and they ask what the product does by itself rather than reporting a result.
- How does Grafana auto-resolve incidents?1
- Do traces reach Grafana?1
- Can logs go to my own Grafana?1
- Does Watchers learn a baseline?1
See the 4 postsHide the posts
do you know how grafana actually auto resolve the incident ?
Native GenAI spans for tool + MCP calls finally make agent latency visible next to the rest of the stack. Are the traces already flowing into your existing Grafana/Datadog or still console-only?
这个功能很酷。不过 Telegram 作为监控通道，如果 Agent 输出日志很频繁，会不会触发 Bot 的速率限制？有没有人试过直接 Webhook 推到本地 Grafana 或 Slack？
Does Watchers learn what normal looks like, or just follow rules?
Wishes· 4 posts· written 27 Oct
What do people want from Grafana Labs?
The 4 wishes are single asks, and two of them are about agents: panels that follow a request through traces and alerts, and one span per tool call carrying the model's intent.
One person wants one span per tool call with the model's stated intent attached, so what an agent planned can be diffed against what it ran. Another wants panels tied to traces and alerts, because charts pulled to the maximum still leave a slow feeling to be chased through the request path.
The other two are about seeing what is already there: one wants the OpenTelemetry data already being collected to be viewable in Grafana. One, writing about a different product, wants a free-form panel UI in the style of Grafana's dashboards, where users define their own panels.
Each comes from one person, so these are individual asks rather than a pattern.
- Flexible user-defined panels1
- Panels linked to traces and alerts1
- Agent spans that show intent1
- See OTel data in Grafana1
See the 4 postsHide the posts
What I'd want from agent traces is one span per tool call with the model's stated intent attached, so you can diff "what it planned" vs "what it ran". That's the most useful signal for debugging coding-agent regressions between eval runs.
grafana面板都拉满了，慢感还是得从链路和告警一起查才香
日志有个Open telemetric直接采集了，想看在grafana看
做成 grafana 那种面板吧，可以自己定义
In their words
How do people talk about Grafana Labs?
The phrases people use most, counted once per person.
- 1“prometheus grafana”4
- 2“monitoring stack”2
- 3“observability tools”2
- 4“ai agents”2
- 5“mcp calls”2
- 6“tool call”2
- 7“give the agent”2
Posts are shown the way X shows them. One of them yours? Here’s why it’s shown, and how to take it off.
Also in Observability and error monitoring
- Datadog+14%The largest group on the page is praise, 12 of 53 people, many of them wiring Datadog alerts and traces into agent workflows; the two complaints raised as often as each other are cost and sensitive data ending up in Datadog.245 posts / 7d
- Sentry−100%Sentry is talked about warmly: plain praise and enthusiasm for its agent and MCP tooling are level at the top of the reception, while the biggest complaint is that self-hosting is heavy and can cost more than the SaaS.198 posts / 7d
Get every post about Grafana Labs, filtered and labelled
This page is the summary. Tracking gives you the stream behind it for any brand, keyword or @mention: each new pain point, request and question as it’s posted, with the spam and promotions already taken out.