fix: address review findings (network ns, journal path, CI semantics)

- network_mode: host so prometheus.exporter.unix reports real host
  interfaces (eth0...) rather than the alloy container's veth pair.
- loki.source.journal: set path = "/var/log/journal" explicitly so it
  doesn't silently fall through to /run/log/journal on volatile-journal
  hosts.
- cadvisor keep-list: add container_memory_working_set_bytes (drives
  several panels on the standard Docker dashboard).
- Drop /dev/kmsg device + extra_hosts:host.docker.internal — neither is
  needed by the current keep-lists, and host-network mode makes the
  extra_hosts entry meaningless.
- CI: extend Alloy validation beyond `fmt` (syntax-only) by booting
  alloy with the embedded config and asserting it stays running, which
  catches bad component refs / wrong arg names that fmt accepts.
- README: refresh Mounts table + Security note to match.
This commit is contained in:
2026-04-25 19:09:49 +07:00
parent 8fd57fe893
commit 638d2f3dd0
3 changed files with 28 additions and 11 deletions
+19
View File
@@ -34,3 +34,22 @@ jobs:
run: |
docker run --rm -v "$PWD/config.alloy:/config.alloy:ro" \
grafana/alloy:v1.10.0 fmt /config.alloy > /dev/null
- name: Validate Alloy config semantics (parse + load components)
# `fmt` is syntax-only; this catches bad component refs, wrong arg
# types, and unknown component types by actually loading the graph.
# Components log scrape errors against the dummy URLs but alloy
# itself stays running, so liveness after sleep == config valid.
run: |
docker run --rm -d --name alloy-test \
-e ALLOY_HOSTNAME -e REMOTECFG_URL -e REMOTECFG_ID -e REMOTECFG_USER \
-e PROM_URL -e PROM_USER -e LOKI_URL -e LOKI_USER -e GRAFANA_TOKEN \
-v "$PWD/config.alloy:/etc/alloy/config.alloy:ro" \
grafana/alloy:v1.10.0 \
run --server.http.listen-addr=127.0.0.1:0 /etc/alloy/config.alloy
sleep 8
if [ "$(docker inspect -f '{{.State.Running}}' alloy-test 2>/dev/null)" != "true" ]; then
docker logs alloy-test
exit 1
fi
docker stop alloy-test >/dev/null
+6 -6
View File
@@ -45,20 +45,20 @@ Same compose file on every host — change `ALLOY_HOSTNAME` and `REMOTECFG_ID` p
## Security note
Runs as `privileged: true` (matching the upstream Grafana Cloud docker integration). This is required for cadvisor to read cgroups via `/sys` and for `/dev/kmsg` access. If you need least-privilege, see the upstream Alloy docker integration docs and tighten capabilities.
Runs `privileged: true` + `network_mode: host`, matching the upstream Grafana Cloud docker integration. `network_mode: host` is required so `prometheus.exporter.unix` reports the host's real network interfaces (eth0…) instead of the alloy container's veth pair. If you need least-privilege, see the upstream Alloy docker integration docs and tighten capabilities.
## Mounts
| Mount | Why |
|---|---|
| `/proc:/rootproc:ro` | node-exporter cpu/mem/load |
| `/proc:/rootproc:ro` | node-exporter cpu/mem/load (referenced via `procfs_path`) |
| `/sys:/sys:ro` | node-exporter + cadvisor cgroups |
| `/:/rootfs:ro` | filesystem collector |
| `/dev/disk/:/dev/disk:ro` | diskstats device labels |
| `/var/run/docker.sock` | docker discovery + log streaming |
| `/:/rootfs:ro` | filesystem collector (referenced via `rootfs_path`) |
| `/dev/disk/:/dev/disk:ro` | node-exporter diskstats device labels |
| `/var/run/docker.sock` | `discovery.docker` + `loki.source.docker` |
| `/var/lib/docker:ro` | cadvisor container metadata |
| `/var/log/journal:ro` | `loki.source.journal` |
| `/dev/kmsg` (device) | cadvisor OOM detection |
| `/etc/machine-id:ro` | stable host id for the journal reader |
| `alloy-data` (named volume) | WAL + remotecfg cache |
## License
+3 -5
View File
@@ -9,6 +9,7 @@ services:
restart: unless-stopped
hostname: ${ALLOY_HOSTNAME:?required}
privileged: true
network_mode: host # node_exporter netdev/netstat sees real host interfaces (eth0…) not veth
environment:
ALLOY_DEPLOY_MODE: docker
REMOTECFG_URL: ${REMOTECFG_URL:?required}
@@ -29,10 +30,6 @@ services:
- /var/lib/docker/:/var/lib/docker:ro
- /var/log/journal:/var/log/journal:ro
- /etc/machine-id:/etc/machine-id:ro # stable host id for systemd journal reader
extra_hosts:
- 'host.docker.internal:host-gateway'
devices:
- /dev/kmsg
configs:
- { source: alloy_config, target: /etc/alloy/config.alloy }
command: 'run --storage.path=/var/lib/alloy/data /etc/alloy/config.alloy'
@@ -151,6 +148,7 @@ configs:
loki.source.journal "default" {
max_age = "12h0m0s"
path = "/var/log/journal"
forward_to = [loki.process.default.receiver]
relabel_rules = loki.relabel.default.rules
}
@@ -201,7 +199,7 @@ configs:
rule {
source_labels = ["__name__"]
regex = "container_cpu_usage_seconds_total|container_fs_reads_total|container_fs_usage_bytes|container_fs_writes_total|container_last_seen|container_memory_usage_bytes|container_network_receive_bytes_total|container_network_receive_errors_total|container_network_receive_packets_dropped_total|container_network_transmit_bytes_total|container_network_transmit_errors_total|container_network_transmit_packets_dropped_total|container_spec_memory_reservation_limit_bytes|machine_memory_bytes|machine_scrape_error|up"
regex = "container_cpu_usage_seconds_total|container_fs_reads_total|container_fs_usage_bytes|container_fs_writes_total|container_last_seen|container_memory_usage_bytes|container_memory_working_set_bytes|container_network_receive_bytes_total|container_network_receive_errors_total|container_network_receive_packets_dropped_total|container_network_transmit_bytes_total|container_network_transmit_errors_total|container_network_transmit_packets_dropped_total|container_spec_memory_reservation_limit_bytes|machine_memory_bytes|machine_scrape_error|up"
action = "keep"
}
}