Running without credentials and going it uses the IAM Profile: ❌

Adding role_arn of said IAM Profile: ❌

Setting up an IAM user with just EC2 read only access and using AWS keys: ✅

That's just fantastic security, #VictoriaMetrics / #Prometheus 🤮

Does anyone use Prometheus/VictoriaMetrics vmagent and (AWS) EC2 Service Discovery with your PromScrape config? I am running it on a server with an IAM Profile that has read-only access to EC2 but:

skipping some ec2_sd_config targets for job_name=ec2_instances because of error: error when fetching instances data from EC2: cannot obtain instances: unexpected status code for "https://ec2.eu-west-2.amazonaws.com/?Action=DescribeInstances&Version=2016-11-15"; got 401; want 200;

I guess it doesn't know how to use an IAM Profile?

#VictoriaMetrics #Prometheus #EC2

Comme il fait gris et que j'ai la flemme, j'ai fait un dashboard Grafana pour Jellyfin 🎬

#Jellyfin #Grafana #Prometheus #SelfHosting

Alt...

Dashboard Grafana sombre pour Jellyfin avec résumé serveur, sessions actives, transcodages, trafic réseau et timeline des médias lus par utilisateur.

#Proxmox #PVE #monitoring #metrics #opensource

https://gyptazy.com/blog/grafana-dashboard-for-pegaprox-prometheus-exporter/

Anomaly detection isn't just for metrics.

#VictoriaMetrics Anomaly Detection supports additional input data sources through #VictoriaLogs reader, allowing you to monitor #log-derived and traces-derived metrics for anomalies.
This expands the versatility, enabling it to handle a wider range of data sources beyond #timeseries metrics from VictoriaMetrics or #Prometheus, including VictoriaLogs and #VictoriaTraces.

@bagder #prometheus @Prometheus is amongst them

RE: https://en.osm.town/@mdione/116239489362801981

OK, the push for AI is official. I'm definitely open for new positions.

I'm a jack of all trades, systems dev and admin, #Linux, #Python, #postgres, #ansible, #prometheus, #grafana, security conscious, I like to review code, and for my hobbies I do some #GIS and would like to do more.

Check my CV https://www.grulic.org.ar/~mdione/

and my blog https://www.grulic.org.ar/~mdione/glob/ (not a typo)

for more details.

#GetFediHired

Marcos Dione boosted

Marcos Dione » 🌐 2026-03-16 / 2026-03-27
@mdione@en.osm.town

I have been suggested that the company I'm working is not doing well. Than and a push for AI is telling me it's maybe time to move on.

What am I? Currently I'm doing #Python, #SysAdmin, GitHub/GitLab admin (yes, both), some Kafka, used to do some Postgres (now we have a DB team), Ansible and more.

Edit: more info: https://en.osm.town/@mdione/116302423501761728

What do I want? No AI, no gambling, no blockchain. In #Marseille or remote. I can invoice abroad, but I would rather be an employee.

#FediHire #job

1/

The Prometheus instance in my homelab just blew up. I need to re boot the frozen host, quickly stop the service then reduce the retention again.

#prometheus #homelab #selfhosting

@amapanda I was using #grok, but it's 6y unmaintained. To me #Loki is not an option, because when I tried to implement #awstats like stats in #prometheus (or loki) + grafana, the problem was the cardinality explosion based on path (even if you just bundle them by app or verb), or on client IP, or on user agent string (not that it makes any sense these days).

I'm thinking of writing a python script to export logs to a SQL db and use #grafana directly to do the queries against that.

Ended up going for a combination of Victoria Metrics/Grafana/AlertManager and other tools that belong to that ecosystem. I've used prometheus often enough, but always found prometheus to be too chonky for my homelab setup, but VM promised less RAM use, and it delivered. So now I'm building grafana dashboards and much more advanced alerting/monitoring than Uptime Kuma could ever do. Bonus points for so many apps having prometheus stats endpoints, which means I now have dashboards for stuff like this GoToSocial instance and much more.

All in all, a much better setup than Uptime Kuma could bring, even if it is more complicated to setup.

@homelab #HomeLab #SelfHosted #Monitoring #Prometheus #VictoriaMetrics #UptimeKuma