Tired of predatory SaaS logging bills charging per GB?
Don't fall for the "zero-cost" myth. You need an enterprise-grade architecture. Our SRE playbook for Ubuntu 24.04 is live:
ā
Write Ahead Logs (WAL) for crash resilience
ā
Grafana Alloy journald socket scraping
ā
Nginx rate limiting for ingestion floods
ā
Fail2ban brute-force mitigation
Deploy on ServerMO Bare Metal to lock in your costs permanently.
I'm gonna try logward and comparing to my loki experience.
š”ļø Open-source, self-hosted log management. Privacy-first alternative to Datadog & ELK. Lightweight, GDPR-compliant, and deployed in minutes with Docker. - logward-dev/logward
And I really appreciate Loki here: I searched through the last 30 days worth of ALL Ceph logs, because the Ceph warning doesn't show when it originally happened. And Loki handled that without any problem at all. Sure, it took a little bit, but it's also running on a Pi CM4. Topped out at 400 MB memory usage, which I'd find pretty acceptable.
#GrafanaLoki v3.4 is here with significant efficiency and standardization improvements!
Key highlight: the integration of the Thanos Object Storage Client, bringing Loki's storage configuration in line with Grafana Mimir & Pyroscope.
Dive deeper into the updates on #InfoQ š https://bit.ly/3FsbNV8
What I did in the last couple of weeks (part 5):
I started reading "Observability with Grafana" by Rob Chapman and Peter Holmes.
https://www.packtpub.com/en-US/product/observability-with-grafana-9781803249643
A nice read so far. They are using Grafana Cloud to make it easy for the reader to get going without having to set up their own observability stack.
They have example code to deploy into a kubernetes cluster (whatever flavor, minikube, ...) so it sends the metrics, logs and traces to Grafana Cloud.
They explain the basics of PromQL, TraceQL and LogQL, the basic architecture of Prometheus, Mimir, Loki and Tempo. They explain OpenTelemetry, they mention Jaeger and Zipkin.
This far a very nice read!
#kubernetes #k8s #observability #grafana #GrafanaLoki #GrafanaTempo #Prometheus #GrafanaMimir #OpenTelemetry
Soak in the Loki knowledge from the 'Recommended Practices for Logging with Grafana Loki' ā”ļø Enlightning stream last Friday!
It is a lighboard collaboration between @nicole, the live audience, and myself! Go team! ā«
ā«~We're going live NOW!
Join @nicole and me for ā”ļø Enlightning and learn about Grafana Loki, an open source logging stack that stores logs in a lightweight yet accessible way ā«
Join us and ask your questions!
youtu.be/p-FFYOM8Y_8
Join ā”ļø Enlightning this Friday, 8/2 and learn Grafana Loki, a set of open source components that can be composed into a fully featured logging stack! @nicole will teach how Loki stores logs in a lightweight yet accessible way ā«
Click 'notify me':
youtu.be/Ep1_zha9ylU
I already know about the line_format function, but the issue is: I'm showing logs from e.g. an entire namespace at the same time. So I can't assume that all the log lines have the same keys after logfmt.
That's why I'd like a function to exclude certain common log line keys from getting displayed, but I wasn't able to do it.
Does anybody have any good idea?
2/2
ServerMO
Adhidarma Hadiwinoto 
Michael
InfoQ
Johannes Kastl
Stefan Gofferje
wiggitywhitney