[serve]
interval_secs = 300
prune_days = 30
listen = "0.0.0.0:9100"
peer_heartbeat_secs = 60
route_check_interval_secs = 300
dashboard = false
# api_token loaded from POM_API_TOKEN env var

[instance]
name = "hetzner"

[targets.mnw]
label = "Makenotwork Production"
expected_routes = ["/", "/discover", "/login", "/docs"]

[[targets.mnw.dns]]
name = "makenot.work"
record_type = "A"
expected = []

[[targets.mnw.dns]]
name = "forums.makenot.work"
record_type = "A"
expected = []

[[targets.mnw.dns]]
name = "git.makenot.work"
record_type = "A"
expected = []

[[targets.mnw.cors]]
url = "https://fsn1.your-objectstorage.com/makenotwork-files/cors-probe"
origin = "https://makenot.work"
method = "PUT"

[targets.mnw.whois]
domain = "makenot.work"
warn_days = 30

[targets.mnw.health]
url = "https://makenot.work/api/health"
timeout_secs = 10

[targets.mnw.health.expect]
status_code = 200
json_fields = { "status" = "operational", "checks.database" = "true" }

[targets.mnw.health.trending]
baseline_window_hours = 168
spike_threshold = 2.0

[targets.mnw.tls]
host = "makenot.work"

[targets.mnw.tests]
# Present in prod's live /etc/pom/pom.toml and previously missing here, which
# made this file unsafe to deploy: installing it would have silently dropped the
# remote CI check from prod. Keep the two in step.
ssh = "max@100.106.221.39"
command = "/home/max/staging/run-ci.sh"
timeout_secs = 600
staleness_days = 7

[targets.mnw.backups]
# Where the backups actually are, verified on prod 2026-07-29: the nightly job
# writes /var/lib/mnw/backups/makenotwork/makenotwork-<date>.sql.gz. This said
# /opt/makenotwork/backups, a path that does not exist on that host, so the
# check reported the target degraded for a directory it could not find while the
# real backups went unwatched. The app lives at /opt/mnw/current, not
# /opt/makenotwork; the old path looks like a pre-rename leftover.
#
# pom matches files by the "<database>-" or "<database>_" prefix inside this
# directory, so the directory is the per-database one, not its parent.
directory = "/var/lib/mnw/backups/makenotwork"
databases = ["makenotwork"]
max_age_hours = 25
interval_secs = 3600

[targets.mnw.scan_pipeline]
# Polls /admin/uploads/health.json for queue depth, stuck-scan count,
# held backlog, and per-layer error rates. Thresholds per
# scan-pipeline-audit.md § 6. Localhost on the makenotwork port to
# skip the Caddy + Cloudflare path (internal-only signal).
base_url = "http://127.0.0.1:3000"
interval_secs = 300
timeout_secs = 10

[targets.mnw.synckit_fleet]
# Which SyncKit SDK versions are actually syncing. SyncKit is client-side
# only, so there is no deployed version to poll; the server aggregates the
# version off each sync request's User-Agent and this reads that. Authed
# with alerts.alerts_ingest_token, so the check does not spawn without one.
# Informational: an old version in the field never degrades the target, only
# a readout PoM cannot take does. Localhost for the same reason as above.
base_url = "http://127.0.0.1:3000"
window_days = 30
interval_secs = 3600
timeout_secs = 10

[targets.mnw-cli]
label = "MNW CLI SSH Server"

[[targets.mnw-cli.dns]]
name = "cli.makenot.work"
record_type = "A"
expected = []

[targets.mnw-cli.ssh_banner]
host = "127.0.0.1"
port = 22
timeout_secs = 5

[targets.mt]
label = "Multithreaded Forum"
expected_routes = ["/"]

[targets.mt.health]
url = "http://127.0.0.1:3400/api/health"
timeout_secs = 5

[targets.mt.health.expect]
status_code = 200
json_fields = { "status" = "operational", "database" = "true" }

[targets.mt.tls]
host = "forums.makenot.work"

[targets.mt.tests]
ssh = "max@100.106.221.39"
command = "cd /home/max/staging/multithreaded && cargo test --workspace 2>&1"
timeout_secs = 300
staleness_days = 7

[targets.htpy]
label = "htpy.app"

[[targets.htpy.dns]]
name = "htpy.app"
record_type = "A"
expected = []

[targets.htpy.whois]
domain = "htpy.app"
warn_days = 30

[targets.htpy.health]
# htpy-web moved twice and this URL followed neither move: it listens on 3200,
# not 8080, and /archive/<name> now 308s to /viewer/<name>. Checking the old
# address failed on connect, so the target has been reading "unreachable" with an
# open incident since 2026-03-29 while htpy.app itself served fine. Verified
# 2026-07-29: this returns 200 with "htpy" in the body from fw13, astra and
# hetzner, and 3200 is in the tailnet ACL's inter-server port list.
url = "http://100.99.153.68:3200/viewer/S_2"
timeout_secs = 10

[targets.htpy.health.expect]
status_code = 200
body_contains = "htpy"

[targets.htpy.tls]
host = "htpy.app"

[peers.astra]
address = "100.106.221.39:9100"
on_missing = "alert"
# token = "<astra's POM_API_TOKEN value>"

[peers.macbook]
address = "100.100.246.136:9100"
on_missing = "log"
# token = "<macbook's POM_API_TOKEN value>"

[alerts]
# postmark_token loaded from POM_POSTMARK_TOKEN env var
to = "pom-alerts@makenot.work"

# MNW operator-log sink: also pushes alerts to POST {mnw_url}/api/internal/alerts.
# Enable once ALERTS_INGEST_TOKEN is set in MNW prod .env AND the endpoint is
# reachable from this host. Confirm reachability first: /api/internal may be
# restricted at the public edge, in which case use the tailnet URL instead of
# the public one below. The token loads from POM_ALERTS_INGEST_TOKEN env var
# (never commit it here). Both mnw_url and the token must be present, else the
# sink stays disabled.
# mnw_url = "https://makenot.work"