Severity: warning · Fires after: 10m · Rule: central/prometheus/rules/visibility.yml
The visibility-shipper sidecar is not being scraped at all — the container is
gone, crashed, or its metrics port is unreachable. SEO snapshots are not
reaching Wikantik.
absent(up{job="visibility-shipper"} == 1)
absent(... == 1) rather than == 0, so it catches both "scraped and failing"
and "not scraped at all". A dead container publishes no up series for the
staleness rule to work with, which is why this exists as a separate signal.
This is a nightly SEO data pipeline. Losing a night of it is worth a morning
look, not a page. That is also why visibility-shipper is excluded from
ServiceDown, which is critical.
ssh jakefear@docker2 'docker compose -f /opt/jakemon/docker-compose.yml ps visibility-shipper'
ssh jakefear@docker2 'docker compose -f /opt/jakemon/docker-compose.yml logs --tail 100 visibility-shipper'
A common non-failure: the shipper idles by design when
WIKANTIK_INSIGHTS_URL / WIKANTIK_INSIGHTS_TOKEN are unset in .env. It logs
not configured: ... Idling. and stays up. That state does not fire this alert.
bin/deploy-central.sh
Related: IncidentVisibilityShipperStale covers the opposite failure — a container that is up and scraped but silently not shipping anything.