3:15
Metrics or logs: why you need both (with a real failure)
Cultura DevOps
2:52
Start Prometheus, Loki and Grafana with a compose file (and check them)
2:35
How Prometheus collects: why it pulls instead of waiting
2:56
The 4 Prometheus metric types and which one fits each case
2:39
PromQL: rate, sum by and the mistake of averaging averages
2:48
Instrument your application without any metrics library
2:11
Your first Grafana dashboard that helps at 3 in the morning
One Grafana dashboard that works for all your services
2:24
Loki does NOT index your logs (and that's why it's cheap)
2:08
From a spike on the graph to the log line that explains it
2:17
Alerts that really ring (and why yours get ignored)
2:37
Cardinality: I added one label and generated 78,000 series in a day
Alertmanager: grouping and silencing without a hundred emails
2:09
Recording rules: precalculate what's expensive in Prometheus
Machine metrics without touching your code: node_exporter
2:29
What each exporter costs you: my app is 13 series out of 3,586
2:06
LogQL: filter by fields inside the log without indexing them
2:15
Metrics from logs, when you can't touch the application
Grafana as code: I delete the container and it comes back with everything
2:46
The 4 golden signals and an SLO with real numbers
10:50
I wrote Prometheus in Python and Grafana connected without noticing
10:05
Loki vs Elasticsearch: Elasticsearch won both searches
