Promtail has reached EOL on March 2, 2026. Grafana recommends Alloy as its replacement. Alloy can convert the promtail configuration file.
I briefly went through the docs, and it also looks like Alloy could replace cadvisor and node_exporter. Alloy can push metrics and logs to Prometheus and Loki, which would simplify the setup because we would only need to configure the same Prometheus server on every server we want to monitor, instead of the current situation, where you have to add every server to the scrape config in Prometheus.
It also looks like Alloy could discover Docker containers through labels. That would create the possibility for every app capable of exposing metrics to put its metric endpoint in a label, and Alloy will automatically collect those metrics. No need to publicly expose endpoints anymore if Alloy can scrape everything locally.
Promtail has reached EOL on March 2, 2026. Grafana recommends Alloy as its replacement. Alloy can convert the promtail configuration file.
I briefly went through the docs, and it also looks like Alloy could replace cadvisor and node_exporter. Alloy can push metrics and logs to Prometheus and Loki, which would simplify the setup because we would only need to configure the same Prometheus server on every server we want to monitor, instead of the current situation, where you have to add every server to the scrape config in Prometheus.
It also looks like Alloy could discover Docker containers through labels. That would create the possibility for every app capable of exposing metrics to put its metric endpoint in a label, and Alloy will automatically collect those metrics. No need to publicly expose endpoints anymore if Alloy can scrape everything locally.
Alloy looks good to me and I'm all for removing moving parts! And if we could collect metrics without exposing them to the world that would be so awesome!!
Alloy looks good to me and I'm all for removing moving parts! And if we could collect metrics without exposing them to the world that would be so awesome!!
> It also looks like Alloy could discover Docker containers through labels
Are you refering to something like this?
https://vincent.bernat.ch/en/blog/2026-prometheus-metrics-discovery-docker-labels
Yes! I think I read another source, but it looks similar.
> > It also looks like Alloy could discover Docker containers through labels
>
> Are you refering to something like this?
>
> https://vincent.bernat.ch/en/blog/2026-prometheus-metrics-discovery-docker-labels
Yes! I think I read another source, but it looks similar.
fauno
self-assigned this 2026-06-10 13:26:30 +00:00
i was looking into doing this for garage, but how would we expose metrics with authorization tokens for instance?
also i think this guide at least assumes the metrics ports will be open to the network, so alloy could reach them on the proxy networks (i'm not sure right now if it's currently possible)
i was looking into doing this for garage, but how would we expose metrics with authorization tokens for instance?
also i think this guide at least assumes the metrics ports will be open to the network, so alloy could reach them on the proxy networks (i'm not sure right now if it's currently possible)
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Promtail has reached EOL on March 2, 2026. Grafana recommends Alloy as its replacement. Alloy can convert the promtail configuration file.
I briefly went through the docs, and it also looks like Alloy could replace cadvisor and node_exporter. Alloy can push metrics and logs to Prometheus and Loki, which would simplify the setup because we would only need to configure the same Prometheus server on every server we want to monitor, instead of the current situation, where you have to add every server to the scrape config in Prometheus.
It also looks like Alloy could discover Docker containers through labels. That would create the possibility for every app capable of exposing metrics to put its metric endpoint in a label, and Alloy will automatically collect those metrics. No need to publicly expose endpoints anymore if Alloy can scrape everything locally.
Alloy looks good to me and I'm all for removing moving parts! And if we could collect metrics without exposing them to the world that would be so awesome!!
Are you refering to something like this?
https://vincent.bernat.ch/en/blog/2026-prometheus-metrics-discovery-docker-labels
Yes! I think I read another source, but it looks similar.
i was looking into doing this for garage, but how would we expose metrics with authorization tokens for instance?
also i think this guide at least assumes the metrics ports will be open to the network, so alloy could reach them on the proxy networks (i'm not sure right now if it's currently possible)
metrics discovery is working at #24