You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Only the monitoring host runs Alloy. morpheus reaches Loki over network
syslog rather than an agent, so it has logs and no host metrics; Saruman and oracle report nothing at all.
Confirmed live: Prometheus has one *-metrics job (prometheus-metrics), and
Loki has two host values (prometheus, morpheus).
config.alloy is written to run unchanged on every host — LOKI_URL and PROMETHEUS_REMOTE_WRITE_URL come from the environment, and
after the fix in #62 so does ALLOY_HOSTNAME — so this is deployment, not
configuration.
Saruman first: it is the Proxmox host, and it is the machine ADR-0007 wants
to build stacks/lab/ on. Deciding what oracle is for is a separate item.
Note docs/architecture.md:134-135 and docs/hardware.md:33 disagree about
how many agents are running; whichever is right, it should match after this.
Tracked in docs/roadmap.md; filed as an issue so it has a place to be discussed and closed.
Activity
fix(observability): narrow to Alertmanager — oracle is a live client (#…
feat(alerts): catch a remote-writing agent that stops pushing